[nodes] Add cancelation to the API

- 86932469e76f1315ee18bfa2fc52b588241dace1 add image_to_dataURL util
- 0c2611059711b45bb6142d30b1d1343ac24268f3 make fast latents method
static
- this method doesn't really need `self` and should be able to be called
without instantiating `Generator`
- 2360bfb6558ea511e9c9576f3d4b5535870d84b4 fix schema gen for
GraphExecutionState
- `GraphExecutionState` uses `default_factory` in its fields; the result
is the OpenAPI schema marks those fields as optional, which propagates
to the generated API client, which means we need a lot of unnecessary
type guards to use this data type. the [simple
fix](https://github.com/pydantic/pydantic/discussions/4577) is to add
config to explicitly say all class properties are required. looks this
this will be resolved in a future pydantic release
- 3cd7319cfdb0f07c6bb12d62d7d02efe1ab12675 fix step callback and fast
latent generation on nodes. have this working in UI. depends on the
small change in #2957

.coveragerc

@@ -0,0 +1,6 @@
+[run]
+omit='.env/*'
+source='.'
+
+[report]
+show_missing = true

.dockerignore

@@ -4,22 +4,22 @@
 !ldm
 !pyproject.toml
 
-# Guard against pulling in any models that might exist in the directory tree
-**/*.pt*
-**/*.ckpt
-
-# ignore frontend but whitelist dist
-invokeai/frontend/
-!invokeai/frontend/dist/
+# ignore frontend/web but whitelist dist
+invokeai/frontend/web/
+!invokeai/frontend/web/dist/
 
 # ignore invokeai/assets but whitelist invokeai/assets/web
 invokeai/assets/
 !invokeai/assets/web/
 
+# Guard against pulling in any models that might exist in the directory tree
+**/*.pt*
+**/*.ckpt
+
 # Byte-compiled / optimized / DLL files
 **/__pycache__/
 **/*.py[cod]
 
 # Distribution / packaging
-*.egg-info/
-*.egg
+**/*.egg-info/
+**/*.egg

.editorconfig

@@ -1,8 +1,5 @@
-root = true
-
 # All files
 [*]
-max_line_length = 80
 charset = utf-8
 end_of_line = lf
 indent_size = 2
@@ -13,18 +10,3 @@ trim_trailing_whitespace = true
 # Python
 [*.py]
 indent_size = 4
-max_line_length = 120
-
-# css
-[*.css]
-indent_size = 4
-
-# flake8
-[.flake8]
-indent_size = 4
-
-# Markdown MkDocs
-[docs/**/*.md]
-max_line_length = 80
-indent_size = 4
-indent_style = unset

.flake8

@@ -1,37 +0,0 @@
-[flake8]
-max-line-length = 120
-extend-ignore =
-    # See https://github.com/PyCQA/pycodestyle/issues/373
-    E203,
-    # use Bugbear's B950 instead
-    E501,
-    # from black repo https://github.com/psf/black/blob/main/.flake8
-    E266, W503, B907
-extend-select =
-    # Bugbear line length
-    B950
-extend-exclude =
-    scripts/orig_scripts/*
-    ldm/models/*
-    ldm/modules/*
-    ldm/data/*
-    ldm/generate.py
-    ldm/util.py
-    ldm/simplet2i.py
-per-file-ignores =
-    # B950 line too long
-    # W605 invalid escape sequence
-    # F841 assigned to but never used
-    # F401 imported but unused
-    tests/test_prompt_parser.py: B950, W605, F401
-    tests/test_textual_inversion.py: F841, B950
-    # B023 Function definition does not bind loop variable
-    scripts/legacy_api.py: F401, B950, B023, F841
-    ldm/invoke/__init__.py: F401
-    # B010 Do not call setattr with a constant attribute value
-    ldm/invoke/server_legacy.py: B010
-# =====================
-# flake-quote settings:
-# =====================
-# Set this to match black style:
-inline-quotes = double

.git-blame-ignore-revs

@@ -0,0 +1 @@
+b3dccfaeb636599c02effc377cdd8a87d658256c

.github/CODEOWNERS

@@ -1,61 +1,34 @@
 # continuous integration
-/.github/workflows/ @mauwii @lstein @blessedcoolant
+/.github/workflows/ @mauwii @lstein
 
 # documentation
-/docs/ @lstein @mauwii @blessedcoolant
-mkdocs.yml @mauwii @lstein
+/docs/ @lstein @mauwii @tildebyte
+/mkdocs.yml @lstein @mauwii
+
+# nodes
+/invokeai/app/ @Kyle0654 @blessedcoolant
 
 # installation and configuration
-/pyproject.toml @mauwii @lstein @ebr
-/docker/ @mauwii
-/scripts/ @ebr @lstein @blessedcoolant
-/installer/ @ebr @lstein 
-ldm/invoke/config @lstein @ebr
-invokeai/assets @lstein @blessedcoolant
-invokeai/configs @lstein @ebr  @blessedcoolant
-/ldm/invoke/_version.py @lstein @blessedcoolant
+/pyproject.toml @mauwii @lstein @blessedcoolant
+/docker/ @mauwii @lstein
+/scripts/ @ebr @lstein
+/installer/ @lstein @ebr
+/invokeai/assets @lstein @ebr
+/invokeai/configs @lstein
+/invokeai/version @lstein @blessedcoolant
 
 # web ui
-/invokeai/frontend @blessedcoolant @psychedelicious
-/invokeai/backend @blessedcoolant @psychedelicious
+/invokeai/frontend @blessedcoolant @psychedelicious @lstein
+/invokeai/backend @blessedcoolant @psychedelicious @lstein
 
-# generation and model management
-/ldm/*.py @lstein @blessedcoolant
-/ldm/generate.py @lstein @keturn
-/ldm/invoke/args.py @lstein @blessedcoolant
-/ldm/invoke/ckpt* @lstein @blessedcoolant
-/ldm/invoke/ckpt_generator @lstein @blessedcoolant
-/ldm/invoke/CLI.py @lstein @blessedcoolant
-/ldm/invoke/config @lstein @ebr @mauwii @blessedcoolant
-/ldm/invoke/generator @keturn @damian0815
-/ldm/invoke/globals.py @lstein @blessedcoolant 
-/ldm/invoke/merge_diffusers.py @lstein @blessedcoolant
-/ldm/invoke/model_manager.py @lstein @blessedcoolant
-/ldm/invoke/txt2mask.py @lstein @blessedcoolant
-/ldm/invoke/patchmatch.py @Kyle0654 @lstein
-/ldm/invoke/restoration @lstein @blessedcoolant
+# generation, model management, postprocessing
+/invokeai/backend @keturn @damian0815 @lstein @blessedcoolant @jpphoto
 
-# attention, textual inversion, model configuration
-/ldm/models @damian0815 @keturn @blessedcoolant
-/ldm/modules/textual_inversion_manager.py @lstein @blessedcoolant
-/ldm/modules/attention.py @damian0815 @keturn
-/ldm/modules/diffusionmodules @damian0815 @keturn
-/ldm/modules/distributions @damian0815 @keturn
-/ldm/modules/ema.py  @damian0815 @keturn
-/ldm/modules/embedding_manager.py @lstein
-/ldm/modules/encoders @damian0815 @keturn
-/ldm/modules/image_degradation @damian0815 @keturn
-/ldm/modules/losses  @damian0815 @keturn
-/ldm/modules/x_transformer.py @damian0815 @keturn
-
-# Nodes
-apps/ @Kyle0654 @jpphoto
-
-# legacy REST API
-# these are dead code
-#/ldm/invoke/pngwriter.py @CapableWeb
-#/ldm/invoke/server_legacy.py @CapableWeb
-#/scripts/legacy_api.py @CapableWeb
-#/tests/legacy_tests.sh @CapableWeb
+# front ends
+/invokeai/frontend/CLI @lstein
+/invokeai/frontend/install @lstein @ebr @mauwii 
+/invokeai/frontend/merge @lstein @blessedcoolant @hipsterusername
+/invokeai/frontend/training @lstein @blessedcoolant @hipsterusername
+/invokeai/frontend/web @psychedelicious @blessedcoolant
 
 

.github/ISSUE_TEMPLATE/BUG_REPORT.yml

@@ -65,6 +65,16 @@ body:
       placeholder: 8GB
     validations:
       required: false
+      
+  - type: input
+    id: version-number
+    attributes:
+      label: What version did you experience this issue on?
+      description: |
+        Please share the version of Invoke AI that you experienced the issue on. If this is not the latest version, please update first to confirm the issue still exists. If you are testing main, please include the commit hash instead.
+      placeholder: X.X.X
+    validations:
+      required: true
 
   - type: textarea
     id: what-happened

.github/workflows/build-container.yml

@@ -5,18 +5,17 @@ on:
       - 'main'
       - 'update/ci/docker/*'
       - 'update/docker/*'
+      - 'dev/ci/docker/*'
+      - 'dev/docker/*'
     paths:
       - 'pyproject.toml'
-      - 'ldm/**'
-      - 'invokeai/backend/**'
-      - 'invokeai/configs/**'
-      - 'invokeai/frontend/dist/**'
+      - '.dockerignore'
+      - 'invokeai/**'
       - 'docker/Dockerfile'
     tags:
       - 'v*.*.*'
   workflow_dispatch:
 
-
 jobs:
   docker:
     if: github.event.pull_request.draft == false
@@ -24,11 +23,11 @@ jobs:
       fail-fast: false
       matrix:
         flavor:
-          - amd
+          - rocm
           - cuda
           - cpu
         include:
-          - flavor: amd
+          - flavor: rocm
             pip-extra-index-url: 'https://download.pytorch.org/whl/rocm5.2'
           - flavor: cuda
             pip-extra-index-url: ''
@@ -54,9 +53,9 @@ jobs:
           tags: |
             type=ref,event=branch
             type=ref,event=tag
-            type=semver,pattern={{version}}
-            type=semver,pattern={{major}}.{{minor}}
-            type=semver,pattern={{major}}
+            type=pep440,pattern={{version}}
+            type=pep440,pattern={{major}}.{{minor}}
+            type=pep440,pattern={{major}}
             type=sha,enable=true,prefix=sha-,format=short
           flavor: |
             latest=${{ matrix.flavor == 'cuda' && github.ref == 'refs/heads/main' }}
@@ -92,7 +91,7 @@ jobs:
           context: .
           file: ${{ env.DOCKERFILE }}
           platforms: ${{ env.PLATFORMS }}
-          push: ${{ github.ref == 'refs/heads/main' || github.ref == 'refs/tags/*' }}
+          push: ${{ github.ref == 'refs/heads/main' || github.ref_type == 'tag' }}
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}
           build-args: PIP_EXTRA_INDEX_URL=${{ matrix.pip-extra-index-url }}

.github/workflows/close-inactive-issues.yml

@@ -0,0 +1,27 @@
+name: Close inactive issues
+on:
+  schedule:
+    - cron: "00 6 * * *"
+
+env:
+  DAYS_BEFORE_ISSUE_STALE: 14
+  DAYS_BEFORE_ISSUE_CLOSE: 28
+
+jobs:
+  close-issues:
+    runs-on: ubuntu-latest
+    permissions:
+      issues: write
+      pull-requests: write
+    steps:
+      - uses: actions/stale@v5
+        with:
+          days-before-issue-stale: ${{ env.DAYS_BEFORE_ISSUE_STALE }}
+          days-before-issue-close: ${{ env.DAYS_BEFORE_ISSUE_CLOSE }}
+          stale-issue-label: "Inactive Issue"
+          stale-issue-message: "There has been no activity in this issue for ${{ env.DAYS_BEFORE_ISSUE_STALE }} days. If this issue is still being experienced, please reply with an updated confirmation that the issue is still being experienced with the latest release."
+          close-issue-message: "Due to inactivity, this issue was automatically closed. If you are still experiencing the issue, please recreate the issue."
+          days-before-pr-stale: -1
+          days-before-pr-close: -1
+          repo-token: ${{ secrets.GITHUB_TOKEN }}
+          operations-per-run: 500

.github/workflows/lint-frontend.yml

@@ -3,14 +3,22 @@ name: Lint frontend
 on:
   pull_request:
     paths:
-      - 'invokeai/frontend/**'
+      - 'invokeai/frontend/web/**'
+    types:
+      - 'ready_for_review'
+      - 'opened'
+      - 'synchronize'
   push:
+    branches:
+      - 'main'
     paths:
-      - 'invokeai/frontend/**'
+      - 'invokeai/frontend/web/**'
+  merge_group:
+  workflow_dispatch:
 
 defaults:
   run:
-    working-directory: invokeai/frontend
+    working-directory: invokeai/frontend/web
 
 jobs:
   lint-frontend:
@@ -23,7 +31,7 @@ jobs:
           node-version: '18'
       - uses: actions/checkout@v3
       - run: 'yarn install --frozen-lockfile'
-      - run: 'yarn tsc'
-      - run: 'yarn run madge'
-      - run: 'yarn run lint --max-warnings=0'
-      - run: 'yarn run prettier --check'
+      - run: 'yarn run lint:tsc'
+      - run: 'yarn run lint:madge'
+      - run: 'yarn run lint:eslint'
+      - run: 'yarn run lint:prettier'

.github/workflows/mkdocs-material.yml

@@ -9,10 +9,6 @@ jobs:
   mkdocs-material:
     if: github.event.pull_request.draft == false
     runs-on: ubuntu-latest
-    env:
-      REPO_URL: '${{ github.server_url }}/${{ github.repository }}'
-      REPO_NAME: '${{ github.repository }}'
-      SITE_URL: 'https://${{ github.repository_owner }}.github.io/InvokeAI'
     steps:
       - name: checkout sources
         uses: actions/checkout@v3
@@ -23,15 +19,11 @@ jobs:
         uses: actions/setup-python@v4
         with:
           python-version: '3.10'
-          cache: pip
-          cache-dependency-path: pyproject.toml
 
       - name: install requirements
-        env:
-          PIP_USE_PEP517: 1
         run: |
           python -m \
-            pip install ".[docs]"
+            pip install -r docs/requirements-mkdocs.txt
 
       - name: confirm buildability
         run: |

.github/workflows/pypi-release.yml

@@ -3,7 +3,7 @@ name: PyPI Release
 on:
   push:
     paths:
-      - 'ldm/invoke/_version.py'
+      - 'invokeai/version/invokeai_version.py'
   workflow_dispatch:
 
 jobs:

.github/workflows/test-invoke-pip-skip.yml

@@ -1,12 +1,12 @@
 name: Test invoke.py pip
 on:
   pull_request:
-    paths-ignore:
-      - 'pyproject.toml'
-      - 'ldm/**'
-      - 'invokeai/backend/**'
-      - 'invokeai/configs/**'
-      - 'invokeai/frontend/dist/**'
+    paths:
+      - '**'
+      - '!pyproject.toml'
+      - '!invokeai/**'
+      - 'invokeai/frontend/web/**'
+      - '!invokeai/frontend/web/dist/**'
   merge_group:
   workflow_dispatch:
 

.github/workflows/test-invoke-pip.yml

@@ -5,17 +5,15 @@ on:
       - 'main'
     paths:
       - 'pyproject.toml'
-      - 'ldm/**'
-      - 'invokeai/backend/**'
-      - 'invokeai/configs/**'
-      - 'invokeai/frontend/dist/**'
+      - 'invokeai/**'
+      - '!invokeai/frontend/web/**'
+      - 'invokeai/frontend/web/dist/**'
   pull_request:
     paths:
       - 'pyproject.toml'
-      - 'ldm/**'
-      - 'invokeai/backend/**'
-      - 'invokeai/configs/**'
-      - 'invokeai/frontend/dist/**'
+      - 'invokeai/**'
+      - '!invokeai/frontend/web/**'
+      - 'invokeai/frontend/web/dist/**'
     types:
       - 'ready_for_review'
       - 'opened'
@@ -112,7 +110,7 @@ jobs:
       - name: set INVOKEAI_OUTDIR
         run: >
           python -c
-          "import os;from ldm.invoke.globals import Globals;OUTDIR=os.path.join(Globals.root,str('outputs'));print(f'INVOKEAI_OUTDIR={OUTDIR}')"
+          "import os;from invokeai.backend.globals import Globals;OUTDIR=os.path.join(Globals.root,str('outputs'));print(f'INVOKEAI_OUTDIR={OUTDIR}')"
           >> ${{ matrix.github-env }}
 
       - name: run invokeai-configure

.gitignore

@@ -68,6 +68,7 @@ htmlcov/
 .cache
 nosetests.xml
 coverage.xml
+cov.xml
 *.cover
 *.py,cover
 .hypothesis/
@@ -197,7 +198,7 @@ checkpoints
 .DS_Store
 
 # Let the frontend manage its own gitignore
-!invokeai/frontend/*
+!invokeai/frontend/web/*
 
 # Scratch folder
 .scratch/
@@ -212,11 +213,6 @@ gfpgan/
 # config file (will be created by installer)
 configs/models.yaml
 
-# weights (will be created by installer)
-models/ldm/stable-diffusion-v1/*.ckpt
-models/clipseg
-models/gfpgan
-
 # ignore initfile
 .invokeai
 
@@ -231,6 +227,3 @@ installer/install.bat
 installer/install.sh
 installer/update.bat
 installer/update.sh
-
-# no longer stored in source directory
-models

.pre-commit-config.yaml

@@ -1,41 +0,0 @@
-# See https://pre-commit.com for more information
-# See https://pre-commit.com/hooks.html for more hooks
-repos:
-  - repo: https://github.com/psf/black
-    rev: 23.1.0
-    hooks:
-      - id: black
-
-  - repo: https://github.com/pycqa/isort
-    rev: 5.12.0
-    hooks:
-      - id: isort
-
-  - repo: https://github.com/PyCQA/flake8
-    rev: 6.0.0
-    hooks:
-      - id: flake8
-        additional_dependencies:
-          - flake8-black
-          - flake8-bugbear
-          - flake8-comprehensions
-          - flake8-simplify
-
-  - repo: https://github.com/pre-commit/mirrors-prettier
-    rev: 'v3.0.0-alpha.4'
-    hooks:
-      - id: prettier
-
-  - repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v4.4.0
-    hooks:
-      - id: check-added-large-files
-      - id: check-executables-have-shebangs
-      - id: check-shebang-scripts-are-executable
-      - id: check-merge-conflict
-      - id: check-symlinks
-      - id: check-toml
-      - id: end-of-file-fixer
-      - id: no-commit-to-branch
-        args: ['--branch', 'main']
-      - id: trailing-whitespace

.prettierignore

@@ -1,14 +0,0 @@
-invokeai/frontend/.husky
-invokeai/frontend/patches
-
-# Ignore artifacts:
-build
-coverage
-static
-invokeai/frontend/dist
-
-# Ignore all HTML files:
-*.html
-
-# Ignore deprecated docs
-docs/installation/deprecated_documentation

.prettierrc.yaml

@@ -1,9 +1,9 @@
-embeddedLanguageFormatting: auto
 endOfLine: lf
-singleQuote: true
-semi: true
-trailingComma: es5
+tabWidth: 2
 useTabs: false
+singleQuote: true
+quoteProps: as-needed
+embeddedLanguageFormatting: auto
 overrides:
   - files: '*.md'
     options:
@@ -11,9 +11,3 @@ overrides:
       printWidth: 80
       parser: markdown
       cursorOffset: -1
-  - files: docs/**/*.md
-    options:
-      tabWidth: 4
-  - files: 'invokeai/frontend/public/locales/*.json'
-    options:
-      tabWidth: 4

.pytest.ini

@@ -0,0 +1,5 @@
+[pytest]
+DJANGO_SETTINGS_MODULE = webtas.settings
+; python_files = tests.py test_*.py *_tests.py
+
+addopts = --cov=. --cov-config=.coveragerc --cov-report xml:cov.xml

README.md

@@ -145,7 +145,7 @@ not supported.
     _For Linux with an AMD GPU:_
 
     ```sh
-    pip install InvokeAI --use-pep517 --extra-index-url https://download.pytorch.org/whl/rocm5.4.2
+    pip install InvokeAI --use-pep517 --extra-index-url https://download.pytorch.org/whl/rocm5.2
     ```
 
     _For Macintoshes, either Intel or M1/M2:_

binary_installer/install.bat.in

@@ -0,0 +1,164 @@
+@echo off
+
+@rem This script will install git (if not found on the PATH variable)
+@rem  using micromamba (an 8mb static-linked single-file binary, conda replacement).
+@rem For users who already have git, this step will be skipped.
+
+@rem Next, it'll download the project's source code.
+@rem Then it will download a self-contained, standalone Python and unpack it.
+@rem Finally, it'll create the Python virtual environment and preload the models.
+
+@rem This enables a user to install this project without manually installing git or Python
+
+@rem change to the script's directory
+PUSHD "%~dp0"
+
+set "no_cache_dir=--no-cache-dir"
+if "%1" == "use-cache" (
+    set "no_cache_dir="
+)
+
+echo ***** Installing InvokeAI.. *****
+@rem Config
+set INSTALL_ENV_DIR=%cd%\installer_files\env
+@rem https://mamba.readthedocs.io/en/latest/installation.html
+set MICROMAMBA_DOWNLOAD_URL=https://github.com/cmdr2/stable-diffusion-ui/releases/download/v1.1/micromamba.exe
+set RELEASE_URL=https://github.com/invoke-ai/InvokeAI
+set RELEASE_SOURCEBALL=/archive/refs/heads/main.tar.gz
+set PYTHON_BUILD_STANDALONE_URL=https://github.com/indygreg/python-build-standalone/releases/download
+set PYTHON_BUILD_STANDALONE=20221002/cpython-3.10.7+20221002-x86_64-pc-windows-msvc-shared-install_only.tar.gz
+
+set PACKAGES_TO_INSTALL=
+
+call git --version >.tmp1 2>.tmp2
+if "%ERRORLEVEL%" NEQ "0" set PACKAGES_TO_INSTALL=%PACKAGES_TO_INSTALL% git
+
+@rem Cleanup
+del /q .tmp1 .tmp2
+
+@rem (if necessary) install git into a contained environment
+if "%PACKAGES_TO_INSTALL%" NEQ "" (
+    @rem download micromamba
+    echo ***** Downloading micromamba from %MICROMAMBA_DOWNLOAD_URL% to micromamba.exe *****
+
+    call curl -L "%MICROMAMBA_DOWNLOAD_URL%" > micromamba.exe
+
+    @rem test the mamba binary
+    echo ***** Micromamba version:  *****
+    call micromamba.exe --version
+
+    @rem create the installer env
+    if not exist "%INSTALL_ENV_DIR%" (
+        call micromamba.exe create -y --prefix "%INSTALL_ENV_DIR%"
+    )
+
+    echo ***** Packages to install:%PACKAGES_TO_INSTALL% *****
+
+    call micromamba.exe install -y --prefix "%INSTALL_ENV_DIR%" -c conda-forge %PACKAGES_TO_INSTALL%
+
+    if not exist "%INSTALL_ENV_DIR%" (
+        echo ----- There was a problem while installing "%PACKAGES_TO_INSTALL%" using micromamba. Cannot continue. -----
+        pause
+        exit /b
+    )
+)
+
+del /q micromamba.exe
+
+@rem For 'git' only
+set PATH=%INSTALL_ENV_DIR%\Library\bin;%PATH%
+
+@rem Download/unpack/clean up InvokeAI release sourceball
+set err_msg=----- InvokeAI source download failed -----
+echo Trying to download "%RELEASE_URL%%RELEASE_SOURCEBALL%"
+curl -L %RELEASE_URL%%RELEASE_SOURCEBALL% --output InvokeAI.tgz
+if %errorlevel% neq 0 goto err_exit
+
+set err_msg=----- InvokeAI source unpack failed -----
+tar -zxf InvokeAI.tgz
+if %errorlevel% neq 0 goto err_exit
+
+del /q InvokeAI.tgz
+
+set err_msg=----- InvokeAI source copy failed -----
+cd InvokeAI-*
+xcopy . .. /e /h
+if %errorlevel% neq 0 goto err_exit
+cd ..
+
+@rem cleanup
+for /f %%i in ('dir /b InvokeAI-*') do rd /s /q %%i
+rd /s /q .dev_scripts .github docker-build tests
+del /q requirements.in requirements-mkdocs.txt shell.nix
+
+echo ***** Unpacked InvokeAI source *****
+
+@rem Download/unpack/clean up python-build-standalone
+set err_msg=----- Python download failed -----
+curl -L %PYTHON_BUILD_STANDALONE_URL%/%PYTHON_BUILD_STANDALONE% --output python.tgz
+if %errorlevel% neq 0 goto err_exit
+
+set err_msg=----- Python unpack failed -----
+tar -zxf python.tgz
+if %errorlevel% neq 0 goto err_exit
+
+del /q python.tgz
+
+echo ***** Unpacked python-build-standalone *****
+
+@rem create venv
+set err_msg=----- problem creating venv -----
+.\python\python -E -s -m venv .venv
+if %errorlevel% neq 0 goto err_exit
+call .venv\Scripts\activate.bat
+
+echo ***** Created Python virtual environment *****
+
+@rem Print venv's Python version
+set err_msg=----- problem calling venv's python -----
+echo We're running under
+.venv\Scripts\python --version
+if %errorlevel% neq 0 goto err_exit
+
+set err_msg=----- pip update failed -----
+.venv\Scripts\python -m pip install %no_cache_dir% --no-warn-script-location --upgrade pip wheel
+if %errorlevel% neq 0 goto err_exit
+
+echo ***** Updated pip and wheel *****
+
+set err_msg=----- requirements file copy failed -----
+copy binary_installer\py3.10-windows-x86_64-cuda-reqs.txt requirements.txt
+if %errorlevel% neq 0 goto err_exit
+
+set err_msg=----- main pip install failed -----
+.venv\Scripts\python -m pip install %no_cache_dir% --no-warn-script-location -r requirements.txt
+if %errorlevel% neq 0 goto err_exit
+
+echo ***** Installed Python dependencies *****
+
+set err_msg=----- InvokeAI setup failed -----
+.venv\Scripts\python -m pip install %no_cache_dir% --no-warn-script-location -e .
+if %errorlevel% neq 0 goto err_exit
+
+copy binary_installer\invoke.bat.in .\invoke.bat
+echo ***** Installed invoke launcher script ******
+
+@rem more cleanup
+rd /s /q binary_installer installer_files
+
+@rem preload the models
+call .venv\Scripts\python ldm\invoke\config\invokeai_configure.py
+set err_msg=----- model download clone failed -----
+if %errorlevel% neq 0 goto err_exit
+deactivate
+
+echo ***** Finished downloading models *****
+
+echo All done! Execute the file invoke.bat in this directory to start InvokeAI
+pause
+exit
+
+:err_exit
+    echo %err_msg%
+    pause
+    exit

binary_installer/install.sh.in

@@ -0,0 +1,235 @@
+#!/usr/bin/env bash
+
+# ensure we're in the correct folder in case user's CWD is somewhere else
+scriptdir=$(dirname "$0")
+cd "$scriptdir"
+
+set -euo pipefail
+IFS=$'\n\t'
+
+function _err_exit {
+    if test "$1" -ne 0
+    then
+        echo -e "Error code $1; Error caught was '$2'"
+        read -p "Press any key to exit..."
+        exit
+    fi
+}
+
+# This script will install git (if not found on the PATH variable)
+#  using micromamba (an 8mb static-linked single-file binary, conda replacement).
+# For users who already have git, this step will be skipped.
+
+# Next, it'll download the project's source code.
+# Then it will download a self-contained, standalone Python and unpack it.
+# Finally, it'll create the Python virtual environment and preload the models.
+
+# This enables a user to install this project without manually installing git or Python
+
+echo -e "\n***** Installing InvokeAI into $(pwd)... *****\n"
+
+export no_cache_dir="--no-cache-dir"
+if [ $# -ge 1 ]; then
+    if [ "$1" = "use-cache" ]; then
+        export no_cache_dir=""
+    fi
+fi
+
+
+OS_NAME=$(uname -s)
+case "${OS_NAME}" in
+    Linux*)     OS_NAME="linux";;
+    Darwin*)    OS_NAME="darwin";;
+    *)          echo -e "\n----- Unknown OS: $OS_NAME! This script runs only on Linux or macOS -----\n" && exit
+esac
+
+OS_ARCH=$(uname -m)
+case "${OS_ARCH}" in
+    x86_64*)    ;;
+    arm64*)     ;;
+    *)          echo -e "\n----- Unknown system architecture: $OS_ARCH! This script runs only on x86_64 or arm64 -----\n" && exit
+esac
+
+# https://mamba.readthedocs.io/en/latest/installation.html
+MAMBA_OS_NAME=$OS_NAME
+MAMBA_ARCH=$OS_ARCH
+if [ "$OS_NAME" == "darwin" ]; then
+    MAMBA_OS_NAME="osx"
+fi
+
+if [ "$OS_ARCH" == "linux" ]; then
+    MAMBA_ARCH="aarch64"
+fi
+
+if [ "$OS_ARCH" == "x86_64" ]; then
+    MAMBA_ARCH="64"
+fi
+
+PY_ARCH=$OS_ARCH
+if [ "$OS_ARCH" == "arm64" ]; then
+    PY_ARCH="aarch64"
+fi
+
+# Compute device ('cd' segment of reqs files) detect goes here
+# This needs a ton of work
+# Suggestions:
+#   - lspci
+#   - check $PATH for nvidia-smi, gtt CUDA/GPU version from output
+#   - Surely there's a similar utility for AMD?
+CD="cuda"
+if [ "$OS_NAME" == "darwin" ] && [ "$OS_ARCH" == "arm64" ]; then
+    CD="mps"
+fi
+
+# config
+INSTALL_ENV_DIR="$(pwd)/installer_files/env"
+MICROMAMBA_DOWNLOAD_URL="https://micro.mamba.pm/api/micromamba/${MAMBA_OS_NAME}-${MAMBA_ARCH}/latest"
+RELEASE_URL=https://github.com/invoke-ai/InvokeAI
+RELEASE_SOURCEBALL=/archive/refs/heads/main.tar.gz
+PYTHON_BUILD_STANDALONE_URL=https://github.com/indygreg/python-build-standalone/releases/download
+if [ "$OS_NAME" == "darwin" ]; then
+    PYTHON_BUILD_STANDALONE=20221002/cpython-3.10.7+20221002-${PY_ARCH}-apple-darwin-install_only.tar.gz
+elif [ "$OS_NAME" == "linux" ]; then
+    PYTHON_BUILD_STANDALONE=20221002/cpython-3.10.7+20221002-${PY_ARCH}-unknown-linux-gnu-install_only.tar.gz
+fi
+echo "INSTALLING $RELEASE_SOURCEBALL FROM $RELEASE_URL"
+
+PACKAGES_TO_INSTALL=""
+
+if ! hash "git" &>/dev/null; then PACKAGES_TO_INSTALL="$PACKAGES_TO_INSTALL git"; fi
+
+# (if necessary) install git and conda into a contained environment
+if [ "$PACKAGES_TO_INSTALL" != "" ]; then
+    # download micromamba
+    echo -e "\n***** Downloading micromamba from $MICROMAMBA_DOWNLOAD_URL to micromamba *****\n"
+
+    curl -L "$MICROMAMBA_DOWNLOAD_URL" | tar -xvjO bin/micromamba > micromamba
+
+    chmod u+x ./micromamba
+
+    # test the mamba binary
+    echo -e "\n***** Micromamba version: *****\n"
+    ./micromamba --version
+
+    # create the installer env
+    if [ ! -e "$INSTALL_ENV_DIR" ]; then
+        ./micromamba create -y --prefix "$INSTALL_ENV_DIR"
+    fi
+
+    echo -e "\n***** Packages to install:$PACKAGES_TO_INSTALL *****\n"
+
+    ./micromamba install -y --prefix "$INSTALL_ENV_DIR" -c conda-forge "$PACKAGES_TO_INSTALL"
+
+    if [ ! -e "$INSTALL_ENV_DIR" ]; then
+        echo -e "\n----- There was a problem while initializing micromamba. Cannot continue. -----\n"
+        exit
+    fi
+fi
+
+rm -f micromamba.exe
+
+export PATH="$INSTALL_ENV_DIR/bin:$PATH"
+
+# Download/unpack/clean up InvokeAI release sourceball
+_err_msg="\n----- InvokeAI source download failed -----\n"
+curl -L $RELEASE_URL/$RELEASE_SOURCEBALL --output InvokeAI.tgz
+_err_exit $? _err_msg
+_err_msg="\n----- InvokeAI source unpack failed -----\n"
+tar -zxf InvokeAI.tgz
+_err_exit $? _err_msg
+
+rm -f InvokeAI.tgz
+
+_err_msg="\n----- InvokeAI source copy failed -----\n"
+cd InvokeAI-*
+cp -r . ..
+_err_exit $? _err_msg
+cd ..
+
+# cleanup
+rm -rf InvokeAI-*/
+rm -rf .dev_scripts/ .github/ docker-build/ tests/ requirements.in requirements-mkdocs.txt shell.nix
+
+echo -e "\n***** Unpacked InvokeAI source *****\n"
+
+# Download/unpack/clean up python-build-standalone
+_err_msg="\n----- Python download failed -----\n"
+curl -L $PYTHON_BUILD_STANDALONE_URL/$PYTHON_BUILD_STANDALONE --output python.tgz
+_err_exit $? _err_msg
+_err_msg="\n----- Python unpack failed -----\n"
+tar -zxf python.tgz
+_err_exit $? _err_msg
+
+rm -f python.tgz
+
+echo -e "\n***** Unpacked python-build-standalone *****\n"
+
+# create venv
+_err_msg="\n----- problem creating venv -----\n"
+
+if [ "$OS_NAME" == "darwin" ]; then
+    # patch sysconfig so that extensions can build properly
+    # adapted from https://github.com/cashapp/hermit-packages/commit/fcba384663892f4d9cfb35e8639ff7a28166ee43
+    PYTHON_INSTALL_DIR="$(pwd)/python"
+    SYSCONFIG="$(echo python/lib/python*/_sysconfigdata_*.py)"
+    TMPFILE="$(mktemp)"
+    chmod +w "${SYSCONFIG}"
+    cp "${SYSCONFIG}" "${TMPFILE}"
+    sed "s,'/install,'${PYTHON_INSTALL_DIR},g" "${TMPFILE}" > "${SYSCONFIG}"
+    rm -f "${TMPFILE}"
+fi
+
+./python/bin/python3 -E -s -m venv .venv
+_err_exit $? _err_msg
+source .venv/bin/activate
+
+echo -e "\n***** Created Python virtual environment *****\n"
+
+# Print venv's Python version
+_err_msg="\n----- problem calling venv's python -----\n"
+echo -e "We're running under"
+.venv/bin/python3 --version
+_err_exit $? _err_msg
+
+_err_msg="\n----- pip update failed -----\n"
+.venv/bin/python3 -m pip install $no_cache_dir --no-warn-script-location --upgrade pip
+_err_exit $? _err_msg
+
+echo -e "\n***** Updated pip *****\n"
+
+_err_msg="\n----- requirements file copy failed -----\n"
+cp binary_installer/py3.10-${OS_NAME}-"${OS_ARCH}"-${CD}-reqs.txt requirements.txt
+_err_exit $? _err_msg
+
+_err_msg="\n----- main pip install failed -----\n"
+.venv/bin/python3 -m pip install $no_cache_dir --no-warn-script-location -r requirements.txt
+_err_exit $? _err_msg
+
+echo -e "\n***** Installed Python dependencies *****\n"
+
+_err_msg="\n----- InvokeAI setup failed -----\n"
+.venv/bin/python3 -m pip install $no_cache_dir --no-warn-script-location -e .
+_err_exit $? _err_msg
+
+echo -e "\n***** Installed InvokeAI *****\n"
+
+cp binary_installer/invoke.sh.in ./invoke.sh
+chmod a+rx ./invoke.sh
+echo -e "\n***** Installed invoke launcher script ******\n"
+
+# more cleanup
+rm -rf binary_installer/ installer_files/
+
+# preload the models
+.venv/bin/python3 scripts/configure_invokeai.py
+_err_msg="\n----- model download clone failed -----\n"
+_err_exit $? _err_msg
+deactivate
+
+echo -e "\n***** Finished downloading models *****\n"
+
+echo "All done! Run the command"
+echo "  $scriptdir/invoke.sh"
+echo "to start InvokeAI."
+read -p "Press any key to exit..."
+exit

binary_installer/invoke.bat.in

@@ -0,0 +1,36 @@
+@echo off
+
+PUSHD "%~dp0"
+call .venv\Scripts\activate.bat
+
+echo Do you want to generate images using the
+echo 1. command-line
+echo 2. browser-based UI
+echo OR
+echo 3. open the developer console
+set /p choice="Please enter 1, 2 or 3: "
+if /i "%choice%" == "1" (
+    echo Starting the InvokeAI command-line.
+    .venv\Scripts\python scripts\invoke.py %*
+) else if /i "%choice%" == "2" (
+    echo Starting the InvokeAI browser-based UI.
+    .venv\Scripts\python scripts\invoke.py --web %*
+) else if /i "%choice%" == "3" (
+    echo Developer Console
+    echo Python command is:
+    where python
+    echo Python version is:
+    python --version
+    echo *************************
+    echo You are now in the system shell, with the local InvokeAI Python virtual environment activated,
+    echo so that you can troubleshoot this InvokeAI installation as necessary.
+    echo *************************
+    echo *** Type `exit` to quit this shell and deactivate the Python virtual environment ***
+    call cmd /k
+) else (
+    echo Invalid selection
+    pause
+    exit /b
+)
+
+deactivate

binary_installer/invoke.sh.in

@@ -0,0 +1,46 @@
+#!/usr/bin/env sh
+
+set -eu
+
+. .venv/bin/activate
+
+# set required env var for torch on mac MPS
+ if [ "$(uname -s)" == "Darwin" ]; then
+     export PYTORCH_ENABLE_MPS_FALLBACK=1
+ fi
+
+echo "Do you want to generate images using the"
+echo "1. command-line"
+echo "2. browser-based UI"
+echo "OR"
+echo "3. open the developer console"
+echo "Please enter 1, 2, or 3:"
+read choice
+
+case $choice in
+    1)
+        printf "\nStarting the InvokeAI command-line..\n";
+        .venv/bin/python scripts/invoke.py $*;
+    ;;
+    2)
+        printf "\nStarting the InvokeAI browser-based UI..\n";
+        .venv/bin/python scripts/invoke.py --web $*;
+    ;;
+    3)
+        printf "\nDeveloper Console:\n";
+        printf "Python command is:\n\t";
+        which python;
+        printf "Python version is:\n\t";
+        python --version;
+        echo "*************************"
+        echo "You are now in your user shell ($SHELL) with the local InvokeAI Python virtual environment activated,";
+        echo "so that you can troubleshoot this InvokeAI installation as necessary.";
+        printf "*************************\n"
+        echo "*** Type \`exit\` to quit this shell and deactivate the Python virtual environment *** ";
+        /usr/bin/env "$SHELL";
+    ;;
+    *)
+        echo "Invalid selection";
+        exit
+    ;;
+esac

binary_installer/readme.txt

@@ -0,0 +1,17 @@
+InvokeAI
+
+Project homepage: https://github.com/invoke-ai/InvokeAI
+
+Installation on Windows:
+    NOTE: You might need to enable Windows Long Paths. If you're not sure,
+    then you almost certainly need to. Simply double-click the 'WinLongPathsEnabled.reg'
+    file. Note that you will need to have admin privileges in order to
+    do this.
+
+    Please double-click the 'install.bat' file (while keeping it inside the invokeAI folder).
+
+Installation on Linux and Mac:
+    Please open the terminal, and run './install.sh' (while keeping it inside the invokeAI folder).
+
+After installation, please run the 'invoke.bat' file (on Windows) or 'invoke.sh'
+file (on Linux/Mac) to start InvokeAI.

binary_installer/requirements.in

@@ -0,0 +1,33 @@
+--prefer-binary
+--extra-index-url https://download.pytorch.org/whl/torch_stable.html
+--extra-index-url https://download.pytorch.org/whl/cu116
+--trusted-host https://download.pytorch.org
+accelerate~=0.15
+albumentations
+diffusers[torch]~=0.11
+einops
+eventlet
+flask_cors
+flask_socketio
+flaskwebgui==1.0.3
+getpass_asterisk
+imageio-ffmpeg
+pyreadline3
+realesrgan
+send2trash
+streamlit
+taming-transformers-rom1504
+test-tube
+torch-fidelity
+torch==1.12.1 ; platform_system == 'Darwin'
+torch==1.12.0+cu116 ; platform_system == 'Linux' or platform_system == 'Windows'
+torchvision==0.13.1 ; platform_system == 'Darwin'
+torchvision==0.13.0+cu116 ; platform_system == 'Linux' or platform_system == 'Windows'
+transformers
+picklescan
+https://github.com/openai/CLIP/archive/d50d76daa670286dd6cacf3bcd80b5e4823fc8e1.zip
+https://github.com/invoke-ai/clipseg/archive/1f754751c85d7d4255fa681f4491ff5711c1c288.zip
+https://github.com/invoke-ai/GFPGAN/archive/3f5d2397361199bc4a91c08bb7d80f04d7805615.zip ; platform_system=='Windows'
+https://github.com/invoke-ai/GFPGAN/archive/c796277a1cf77954e5fc0b288d7062d162894248.zip ; platform_system=='Linux' or platform_system=='Darwin'
+https://github.com/Birch-san/k-diffusion/archive/363386981fee88620709cf8f6f2eea167bd6cd74.zip
+https://github.com/invoke-ai/PyPatchMatch/archive/129863937a8ab37f6bbcec327c994c0f932abdbc.zip

docker/Dockerfile

@@ -4,15 +4,15 @@ ARG PYTHON_VERSION=3.9
 ##################
 ##  base image  ##
 ##################
-FROM python:${PYTHON_VERSION}-slim AS python-base
+FROM --platform=${TARGETPLATFORM} python:${PYTHON_VERSION}-slim AS python-base
 
 LABEL org.opencontainers.image.authors="mauwii@outlook.de"
 
-# prepare for buildkit cache
+# Prepare apt for buildkit cache
 RUN rm -f /etc/apt/apt.conf.d/docker-clean \
   && echo 'Binary::apt::APT::Keep-Downloaded-Packages "true";' >/etc/apt/apt.conf.d/keep-cache
 
-# Install necessary packages
+# Install dependencies
 RUN \
   --mount=type=cache,target=/var/cache/apt,sharing=locked \
   --mount=type=cache,target=/var/lib/apt,sharing=locked \
@@ -23,7 +23,7 @@ RUN \
     libglib2.0-0=2.66.* \
     libopencv-dev=4.5.*
 
-# set working directory and env
+# Set working directory and env
 ARG APPDIR=/usr/src
 ARG APPNAME=InvokeAI
 WORKDIR ${APPDIR}
@@ -32,7 +32,7 @@ ENV PATH ${APPDIR}/${APPNAME}/bin:$PATH
 ENV PYTHONDONTWRITEBYTECODE 1
 # Turns off buffering for easier container logging
 ENV PYTHONUNBUFFERED 1
-# don't fall back to legacy build system
+# Don't fall back to legacy build system
 ENV PIP_USE_PEP517=1
 
 #######################
@@ -40,7 +40,7 @@ ENV PIP_USE_PEP517=1
 #######################
 FROM python-base AS pyproject-builder
 
-# Install dependencies
+# Install build dependencies
 RUN \
   --mount=type=cache,target=/var/cache/apt,sharing=locked \
   --mount=type=cache,target=/var/lib/apt,sharing=locked \
@@ -51,26 +51,30 @@ RUN \
     gcc=4:10.2.* \
     python3-dev=3.9.*
 
-# prepare pip for buildkit cache
+# Prepare pip for buildkit cache
 ARG PIP_CACHE_DIR=/var/cache/buildkit/pip
 ENV PIP_CACHE_DIR ${PIP_CACHE_DIR}
 RUN mkdir -p ${PIP_CACHE_DIR}
 
-# create virtual environment
-RUN --mount=type=cache,target=${PIP_CACHE_DIR},sharing=locked \
+# Create virtual environment
+RUN --mount=type=cache,target=${PIP_CACHE_DIR} \
   python3 -m venv "${APPNAME}" \
   --upgrade-deps
 
-# copy sources
-COPY --link . .
-
-# install pyproject.toml
+# Install requirements
+COPY --link pyproject.toml .
+COPY --link invokeai/version/invokeai_version.py invokeai/version/__init__.py invokeai/version/
 ARG PIP_EXTRA_INDEX_URL
 ENV PIP_EXTRA_INDEX_URL ${PIP_EXTRA_INDEX_URL}
-RUN --mount=type=cache,target=${PIP_CACHE_DIR},sharing=locked \
+RUN --mount=type=cache,target=${PIP_CACHE_DIR} \
+  "${APPNAME}"/bin/pip install .
+
+# Install pyproject.toml
+COPY --link . .
+RUN --mount=type=cache,target=${PIP_CACHE_DIR} \
   "${APPNAME}/bin/pip" install .
 
-# build patchmatch
+# Build patchmatch
 RUN python3 -c "from patchmatch import patch_match"
 
 #####################
@@ -86,14 +90,14 @@ RUN useradd \
   -U \
   "${UNAME}"
 
-# create volume directory
+# Create volume directory
 ARG VOLUME_DIR=/data
 RUN mkdir -p "${VOLUME_DIR}" \
-  && chown -R "${UNAME}" "${VOLUME_DIR}"
+  && chown -hR "${UNAME}:${UNAME}" "${VOLUME_DIR}"
 
-# setup runtime environment
-USER ${UNAME}
-COPY --chown=${UNAME} --from=pyproject-builder ${APPDIR}/${APPNAME} ${APPNAME}
+# Setup runtime environment
+USER ${UNAME}:${UNAME}
+COPY --chown=${UNAME}:${UNAME} --from=pyproject-builder ${APPDIR}/${APPNAME} ${APPNAME}
 ENV INVOKEAI_ROOT ${VOLUME_DIR}
 ENV TRANSFORMERS_CACHE ${VOLUME_DIR}/.cache
 ENV INVOKE_MODEL_RECONFIGURE "--yes --default_only"

docker/build.sh

@@ -41,7 +41,7 @@ else
 fi
 
 # Build Container
-DOCKER_BUILDKIT=1 docker build \
+docker build \
     --platform="${PLATFORM:-linux/amd64}" \
     --tag="${CONTAINER_IMAGE:-invokeai}" \
     ${CONTAINER_FLAVOR:+--build-arg="CONTAINER_FLAVOR=${CONTAINER_FLAVOR}"} \

docker/env.sh

@@ -49,3 +49,6 @@ CONTAINER_FLAVOR="${CONTAINER_FLAVOR-cuda}"
 CONTAINER_TAG="${CONTAINER_TAG-"${INVOKEAI_BRANCH##*/}-${CONTAINER_FLAVOR}"}"
 CONTAINER_IMAGE="${CONTAINER_REGISTRY}/${CONTAINER_REPOSITORY}:${CONTAINER_TAG}"
 CONTAINER_IMAGE="${CONTAINER_IMAGE,,}"
+
+# enable docker buildkit
+export DOCKER_BUILDKIT=1

docker/run.sh

@@ -21,10 +21,10 @@ docker run \
   --tty \
   --rm \
   --platform="${PLATFORM}" \
-  --name="${REPOSITORY_NAME,,}" \
-  --hostname="${REPOSITORY_NAME,,}" \
-  --mount=source="${VOLUMENAME}",target=/data \
-  --mount type=bind,source="$(pwd)"/outputs,target=/data/outputs \
+  --name="${REPOSITORY_NAME}" \
+  --hostname="${REPOSITORY_NAME}" \
+  --mount type=volume,volume-driver=local,source="${VOLUMENAME}",target=/data \
+  --mount type=bind,source="$(pwd)"/outputs/,target=/data/outputs/ \
   ${MODELSPATH:+--mount="type=bind,source=${MODELSPATH},target=/data/models"} \
   ${HUGGING_FACE_HUB_TOKEN:+--env="HUGGING_FACE_HUB_TOKEN=${HUGGING_FACE_HUB_TOKEN}"} \
   --publish=9090:9090 \
@@ -32,7 +32,7 @@ docker run \
   ${GPU_FLAGS:+--gpus="${GPU_FLAGS}"} \
   "${CONTAINER_IMAGE}" ${@:+$@}
 
-# Remove Trash folder
+echo -e "\nCleaning trash folder ..."
 for f in outputs/.Trash*; do
   if [ -e "$f" ]; then
     rm -Rf "$f"

docs/.markdownlint.jsonc

@@ -1,5 +0,0 @@
-{
-  "MD046": false,
-  "MD007": false,
-  "MD030": false
-}

docs/contributing/ARCHITECTURE.md

@@ -0,0 +1,93 @@
+# Invoke.AI Architecture
+
+```mermaid
+flowchart TB
+
+  subgraph apps[Applications]
+    webui[WebUI]
+    cli[CLI]
+
+  subgraph webapi[Web API]
+    api[HTTP API]
+    sio[Socket.IO]
+  end
+
+  end
+
+  subgraph invoke[Invoke]
+    direction LR
+    invoker
+    services
+    sessions
+    invocations
+  end
+
+  subgraph core[AI Core]
+    Generate
+  end
+
+  webui --> webapi
+  webapi --> invoke
+  cli --> invoke
+
+  invoker --> services & sessions
+  invocations --> services
+  sessions --> invocations
+
+  services --> core
+
+  %% Styles
+  classDef sg fill:#5028C8,font-weight:bold,stroke-width:2,color:#fff,stroke:#14141A
+  classDef default stroke-width:2px,stroke:#F6B314,color:#fff,fill:#14141A
+
+  class apps,webapi,invoke,core sg
+
+```
+
+## Applications
+
+Applications are built on top of the invoke framework. They should construct `invoker` and then interact through it. They should avoid interacting directly with core code in order to support a variety of configurations.
+
+### Web UI
+
+The Web UI is built on top of an HTTP API built with [FastAPI](https://fastapi.tiangolo.com/) and [Socket.IO](https://socket.io/). The frontend code is found in `/frontend` and the backend code is found in `/ldm/invoke/app/api_app.py` and `/ldm/invoke/app/api/`. The code is further organized as such:
+
+| Component | Description |
+| --- | --- |
+| api_app.py | Sets up the API app, annotates the OpenAPI spec with additional data, and runs the API |
+| dependencies | Creates all invoker services and the invoker, and provides them to the API |
+| events | An eventing system that could in the future be adapted to support horizontal scale-out |
+| sockets | The Socket.IO interface - handles listening to and emitting session events (events are defined in the events service module) |
+| routers | API definitions for different areas of API functionality |
+
+### CLI
+
+The CLI is built automatically from invocation metadata, and also supports invocation piping and auto-linking. Code is available in `/ldm/invoke/app/cli_app.py`.
+
+## Invoke
+
+The Invoke framework provides the interface to the underlying AI systems and is built with flexibility and extensibility in mind. There are four major concepts: invoker, sessions, invocations, and services.
+
+### Invoker
+
+The invoker (`/ldm/invoke/app/services/invoker.py`) is the primary interface through which applications interact with the framework. Its primary purpose is to create, manage, and invoke sessions. It also maintains two sets of services:
+- **invocation services**, which are used by invocations to interact with core functionality.
+- **invoker services**, which are used by the invoker to manage sessions and manage the invocation queue.
+
+### Sessions
+
+Invocations and links between them form a graph, which is maintained in a session. Sessions can be queued for invocation, which will execute their graph (either the next ready invocation, or all invocations). Sessions also maintain execution history for the graph (including storage of any outputs). An invocation may be added to a session at any time, and there is capability to add and entire graph at once, as well as to automatically link new invocations to previous invocations. Invocations can not be deleted or modified once added.
+
+The session graph does not support looping. This is left as an application problem to prevent additional complexity in the graph.
+
+### Invocations
+
+Invocations represent individual units of execution, with inputs and outputs. All invocations are located in `/ldm/invoke/app/invocations`, and are all automatically discovered and made available in the applications. These are the primary way to expose new functionality in Invoke.AI, and the [implementation guide](INVOCATIONS.md) explains how to add new invocations.
+
+### Services
+
+Services provide invocations access AI Core functionality and other necessary functionality (e.g. image storage). These are available in `/ldm/invoke/app/services`. As a general rule, new services should provide an interface as an abstract base class, and may provide a lightweight local implementation by default in their module. The goal for all services should be to enable the usage of different implementations (e.g. using cloud storage for image storage), but should not load any module dependencies unless that implementation has been used (i.e. don't import anything that won't be used, especially if it's expensive to import).
+
+## AI Core
+
+The AI Core is represented by the rest of the code base (i.e. the code outside of `/ldm/invoke/app/`).

docs/contributing/INVOCATIONS.md

@@ -0,0 +1,105 @@
+# Invocations
+
+Invocations represent a single operation, its inputs, and its outputs. These operations and their outputs can be chained together to generate and modify images.
+
+## Creating a new invocation
+
+To create a new invocation, either find the appropriate module file in `/ldm/invoke/app/invocations` to add your invocation to, or create a new one in that folder. All invocations in that folder will be discovered and made available to the CLI and API automatically. Invocations make use of [typing](https://docs.python.org/3/library/typing.html) and [pydantic](https://pydantic-docs.helpmanual.io/) for validation and integration into the CLI and API.
+
+An invocation looks like this:
+
+```py
+class UpscaleInvocation(BaseInvocation):
+    """Upscales an image."""
+    type: Literal['upscale'] = 'upscale'
+
+    # Inputs
+    image: Union[ImageField,None] = Field(description="The input image")
+    strength: float               = Field(default=0.75, gt=0, le=1, description="The strength")
+    level: Literal[2,4]           = Field(default=2, description = "The upscale level")
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = context.services.images.get(self.image.image_type, self.image.image_name)
+        results = context.services.generate.upscale_and_reconstruct(
+            image_list     = [[image, 0]],
+            upscale        = (self.level, self.strength),
+            strength       = 0.0, # GFPGAN strength
+            save_original  = False,
+            image_callback = None,
+        )
+
+        # Results are image and seed, unwrap for now
+        # TODO: can this return multiple results?
+        image_type = ImageType.RESULT
+        image_name = context.services.images.create_name(context.graph_execution_state_id, self.id)
+        context.services.images.save(image_type, image_name, results[0][0])
+        return ImageOutput(
+            image = ImageField(image_type = image_type, image_name = image_name)
+        )
+```
+
+Each portion is important to implement correctly.
+
+### Class definition and type
+```py
+class UpscaleInvocation(BaseInvocation):
+    """Upscales an image."""
+    type: Literal['upscale'] = 'upscale'
+```
+All invocations must derive from `BaseInvocation`. They should have a docstring that declares what they do in a single, short line. They should also have a `type` with a type hint that's `Literal["command_name"]`, where `command_name` is what the user will type on the CLI or use in the API to create this invocation. The `command_name` must be unique. The `type` must be assigned to the value of the literal in the type hint.
+
+### Inputs
+```py
+    # Inputs
+    image: Union[ImageField,None] = Field(description="The input image")
+    strength: float               = Field(default=0.75, gt=0, le=1, description="The strength")
+    level: Literal[2,4]           = Field(default=2, description="The upscale level")
+```
+Inputs consist of three parts: a name, a type hint, and a `Field` with default, description, and validation information. For example:
+| Part | Value | Description |
+| ---- | ----- | ----------- |
+| Name | `strength` | This field is referred to as `strength` |
+| Type Hint | `float` | This field must be of type `float` |
+| Field | `Field(default=0.75, gt=0, le=1, description="The strength")` | The default value is `0.75`, the value must be in the range (0,1], and help text will show "The strength" for this field. |
+
+Notice that `image` has type `Union[ImageField,None]`. The `Union` allows this field to be parsed with `None` as a value, which enables linking to previous invocations. All fields should either provide a default value or allow `None` as a value, so that they can be overwritten with a linked output from another invocation.
+
+The special type `ImageField` is also used here. All images are passed as `ImageField`, which protects them from pydantic validation errors (since images only ever come from links).
+
+Finally, note that for all linking, the `type` of the linked fields must match. If the `name` also matches, then the field can be **automatically linked** to a previous invocation by name and matching.
+
+### Invoke Function
+```py
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = context.services.images.get(self.image.image_type, self.image.image_name)
+        results = context.services.generate.upscale_and_reconstruct(
+            image_list     = [[image, 0]],
+            upscale        = (self.level, self.strength),
+            strength       = 0.0, # GFPGAN strength
+            save_original  = False,
+            image_callback = None,
+        )
+
+        # Results are image and seed, unwrap for now
+        image_type = ImageType.RESULT
+        image_name = context.services.images.create_name(context.graph_execution_state_id, self.id)
+        context.services.images.save(image_type, image_name, results[0][0])
+        return ImageOutput(
+            image = ImageField(image_type = image_type, image_name = image_name)
+        )
+```
+The `invoke` function is the last portion of an invocation. It is provided an `InvocationContext` which contains services to perform work as well as a `session_id` for use as needed. It should return a class with output values that derives from `BaseInvocationOutput`.
+
+Before being called, the invocation will have all of its fields set from defaults, inputs, and finally links (overriding in that order).
+
+Assume that this invocation may be running simultaneously with other invocations, may be running on another machine, or in other interesting scenarios. If you need functionality, please provide it as a service in the `InvocationServices` class, and make sure it can be overridden.
+
+### Outputs
+```py
+class ImageOutput(BaseInvocationOutput):
+    """Base class for invocations that output an image"""
+    type: Literal['image'] = 'image'
+
+    image: ImageField = Field(default=None, description="The output image")
+```
+Output classes look like an invocation class without the invoke method. Prefer to use an existing output class if available, and prefer to name inputs the same as outputs when possible, to promote automatic invocation linking.

docs/features/TEXTUAL_INVERSION.md

@@ -154,11 +154,8 @@ training sets will converge with 2000-3000 steps.
 
 This adjusts how many training images are processed simultaneously in
 each step. Higher values will cause the training process to run more
-quickly, but use more memory. The default size is selected based on
-whether you have the `xformers` memory-efficient attention library
-installed. If `xformers` is available, the batch size will be 8,
-otherwise 3.  These values were chosen to allow training to run with
-GPUs with as little as 12 GB VRAM.
+quickly, but use more memory. The default size will run with GPUs with
+as little as 12 GB.
 
 ### Learning rate
 
@@ -175,10 +172,8 @@ learning rate to improve performance.
 
 ### Use xformers acceleration
 
-This will activate XFormers memory-efficient attention, which will
-reduce memory requirements by half or more and allow you to select a
-higher batch size. You need to have XFormers installed for this to
-have an effect.
+This will activate XFormers memory-efficient attention. You need to
+have XFormers installed for this to have an effect.
 
 ### Learning rate scheduler
 
@@ -255,49 +250,6 @@ invokeai-ti \
        --only_save_embeds
 ```
 
-## Using Distributed Training
-
-If you have multiple GPUs on one machine, or a cluster of GPU-enabled
-machines, you can activate distributed training. See the [HuggingFace
-Accelerate pages](https://huggingface.co/docs/accelerate/index) for
-full information, but the basic recipe is:
-
-1. Enter the InvokeAI developer's console command line by selecting
-option [8] from the `invoke.sh`/`invoke.bat` script.
-
-2. Configurate Accelerate using `accelerate config`:
-```sh
-accelerate config
-```
-This will guide you through the configuration process, including
-specifying how many machines you will run training on and the number
-of GPUs pe rmachine.
-
-You only need to do this once.
-
-3. Launch training from the command line using `accelerate launch`. Be sure
-that your current working directory is the InvokeAI root directory (usually
-named `invokeai` in your home directory):
-
-```sh
-accelerate launch .venv/bin/invokeai-ti \
-       --model=stable-diffusion-1.5 \
-       --resolution=512 \
-       --learnable_property=object \
-       --initializer_token='*' \
-       --placeholder_token='<shraddha>' \
-       --train_data_dir=/home/lstein/invokeai/text-inversion-training-data/shraddha \
-       --output_dir=/home/lstein/invokeai/text-inversion-training/shraddha \
-       --scale_lr \
-       --train_batch_size=10 \
-       --gradient_accumulation_steps=4 \
-       --max_train_steps=2000 \
-       --learning_rate=0.0005 \
-       --lr_scheduler=constant \
-       --mixed_precision=fp16 \
-       --only_save_embeds
-```
-
 ## Using Embeddings
 
 After training completes, the resultant embeddings will be saved into your `$INVOKEAI_ROOT/embeddings/<trigger word>/learned_embeds.bin`.

docs/features/index.md

@@ -2,82 +2,62 @@
 title: Overview
 ---
 
--   The Basics
+Here you can find the documentation for InvokeAI's various features.
 
-    -   The [Web User Interface](WEB.md)
+## The Basics
+### * The [Web User Interface](WEB.md)
+Guide to the Web interface. Also see the [WebUI Hotkeys Reference Guide](WEBUIHOTKEYS.md)
 
-        Guide to the Web interface. Also see the
-        [WebUI Hotkeys Reference Guide](WEBUIHOTKEYS.md)
+### * The [Unified Canvas](UNIFIED_CANVAS.md)
+Build complex scenes by combine and modifying multiple images in a stepwise
+fashion. This feature combines img2img, inpainting and outpainting in
+a single convenient digital artist-optimized user interface.
 
-    -   The [Unified Canvas](UNIFIED_CANVAS.md)
+### * The [Command Line Interface (CLI)](CLI.md)
+Scriptable access to InvokeAI's features.
 
-        Build complex scenes by combine and modifying multiple images in a
-        stepwise fashion. This feature combines img2img, inpainting and
-        outpainting in a single convenient digital artist-optimized user
-        interface.
+## Image Generation
+### * [Prompt Engineering](PROMPTS.md)
+Get the images you want with the InvokeAI  prompt engineering language.
 
-    -   The [Command Line Interface (CLI)](CLI.md)
+## * [Post-Processing](POSTPROCESS.md)
+Restore mangled faces and make images larger with upscaling. Also see the [Embiggen Upscaling Guide](EMBIGGEN.md).
 
-        Scriptable access to InvokeAI's features.
+## * The [Concepts Library](CONCEPTS.md)
+Add custom subjects and styles using HuggingFace's repository of embeddings.
 
--   Image Generation
+### * [Image-to-Image Guide for the CLI](IMG2IMG.md)
+Use a seed image to build new creations in the CLI.
 
-    -   [Prompt Engineering](PROMPTS.md)
+### * [Inpainting Guide for the CLI](INPAINTING.md)
+Selectively erase and replace portions of an existing image in the CLI.
 
-        Get the images you want with the InvokeAI prompt engineering language.
+### * [Outpainting Guide for the CLI](OUTPAINTING.md)
+Extend the borders of the image with an "outcrop" function within the CLI.
 
-    -   [Post-Processing](POSTPROCESS.md)
+### * [Generating Variations](VARIATIONS.md)
+Have an image you like and want to generate many more like it? Variations
+are the ticket.
 
-        Restore mangled faces and make images larger with upscaling. Also see
-        the [Embiggen Upscaling Guide](EMBIGGEN.md).
+## Model Management
 
-    -   The [Concepts Library](CONCEPTS.md)
+## * [Model Installation](../installation/050_INSTALLING_MODELS.md)
+Learn how to import third-party models and switch among them. This
+guide also covers optimizing models to load quickly.
 
-        Add custom subjects and styles using HuggingFace's repository of
-        embeddings.
+## * [Merging Models](MODEL_MERGING.md)
+Teach an old model new tricks. Merge 2-3 models together to create a
+new model that combines characteristics of the originals.
 
-    -   [Image-to-Image Guide for the CLI](IMG2IMG.md)
+## * [Textual Inversion](TEXTUAL_INVERSION.md)
+Personalize models by adding your own style or subjects.
 
-        Use a seed image to build new creations in the CLI.
+# Other Features
 
-    -   [Inpainting Guide for the CLI](INPAINTING.md)
+## * [The NSFW Checker](NSFW.md)
+Prevent InvokeAI from displaying unwanted racy images.
 
-        Selectively erase and replace portions of an existing image in the CLI.
-
-    -   [Outpainting Guide for the CLI](OUTPAINTING.md)
-
-        Extend the borders of the image with an "outcrop" function within the
-        CLI.
-
-    -   [Generating Variations](VARIATIONS.md)
-
-        Have an image you like and want to generate many more like it?
-        Variations are the ticket.
-
--   Model Management
-
-    -   [Model Installation](../installation/050_INSTALLING_MODELS.md)
-
-        Learn how to import third-party models and switch among them. This guide
-        also covers optimizing models to load quickly.
-
-    -   [Merging Models](MODEL_MERGING.md)
-
-        Teach an old model new tricks. Merge 2-3 models together to create a new
-        model that combines characteristics of the originals.
-
-    -   [Textual Inversion](TEXTUAL_INVERSION.md)
-
-        Personalize models by adding your own style or subjects.
-
--   Other Features
-
-    -   [The NSFW Checker](NSFW.md)
-
-        Prevent InvokeAI from displaying unwanted racy images.
-
-    -   [Miscellaneous](OTHER.md)
-
-        Run InvokeAI on Google Colab, generate images with repeating patterns,
-        batch process a file of prompts, increase the "creativity" of image
-        generation by adding initial noise, and more!
+## * [Miscellaneous](OTHER.md)
+Run InvokeAI on Google Colab, generate images with repeating patterns,
+batch process a file of prompts, increase the "creativity" of image
+generation by adding initial noise, and more!

docs/help/IDE-Settings/index.md

@@ -1,4 +0,0 @@
-# :octicons-file-code-16: IDE-Settings
-
-Here we will share settings for IDEs used by our developers, maybe you can find
-something interestening which will help to boost your development efficency 🔥

docs/help/IDE-Settings/vs-code.md

@@ -1,250 +0,0 @@
----
-title: Visual Studio Code
----
-
-# :material-microsoft-visual-studio-code:Visual Studio Code
-
-The Workspace Settings are stored in the project (repository) root and get
-higher priorized than your user settings.
-
-This helps to have different settings for different projects, while the user
-settings get used as a default value if no workspace settings are provided.
-
-## tasks.json
-
-First we will create a task configuration which will create a virtual
-environment and update the deps (pip, setuptools and wheel).
-
-Into this venv we will then install the pyproject.toml in editable mode with
-dev, docs and test dependencies.
-
-```json title=".vscode/tasks.json"
-{
-    // See https://go.microsoft.com/fwlink/?LinkId=733558
-    // for the documentation about the tasks.json format
-    "version": "2.0.0",
-    "tasks": [
-        {
-            "label": "Create virtual environment",
-            "detail": "Create .venv and upgrade pip, setuptools and wheel",
-            "command": "python3",
-            "args": [
-                "-m",
-                "venv",
-                ".venv",
-                "--prompt",
-                "InvokeAI",
-                "--upgrade-deps"
-            ],
-            "runOptions": {
-                "instanceLimit": 1,
-                "reevaluateOnRerun": true
-            },
-            "group": {
-                "kind": "build"
-            },
-            "presentation": {
-                "echo": true,
-                "reveal": "always",
-                "focus": false,
-                "panel": "shared",
-                "showReuseMessage": true,
-                "clear": false
-            }
-        },
-        {
-            "label": "build InvokeAI",
-            "detail": "Build pyproject.toml with extras dev, docs and test",
-            "command": "${workspaceFolder}/.venv/bin/python3",
-            "args": [
-                "-m",
-                "pip",
-                "install",
-                "--use-pep517",
-                "--editable",
-                ".[dev,docs,test]"
-            ],
-            "dependsOn": "Create virtual environment",
-            "dependsOrder": "sequence",
-            "group": {
-                "kind": "build",
-                "isDefault": true
-            },
-            "presentation": {
-                "echo": true,
-                "reveal": "always",
-                "focus": false,
-                "panel": "shared",
-                "showReuseMessage": true,
-                "clear": false
-            }
-        }
-    ]
-}
-```
-
-The fastest way to build InvokeAI now is ++cmd+shift+b++
-
-## launch.json
-
-This file is used to define debugger configurations, so that you can one-click
-launch and monitor the application, set halt points to inspect specific states,
-...
-
-```json title=".vscode/launch.json"
-{
-    "version": "0.2.0",
-    "configurations": [
-        {
-            "name": "invokeai web",
-            "type": "python",
-            "request": "launch",
-            "program": ".venv/bin/invokeai",
-            "justMyCode": true
-        },
-        {
-            "name": "invokeai cli",
-            "type": "python",
-            "request": "launch",
-            "program": ".venv/bin/invokeai",
-            "justMyCode": true
-        },
-        {
-            "name": "mkdocs serve",
-            "type": "python",
-            "request": "launch",
-            "program": ".venv/bin/mkdocs",
-            "args": ["serve"],
-            "justMyCode": true
-        }
-    ]
-}
-```
-
-Then you only need to hit ++f5++ and the fun begins :nerd: (It is asumed that
-you have created a virtual environment via the [tasks](#tasksjson) from the
-previous step.)
-
-## extensions.json
-
-A list of recommended vscode-extensions to make your life easier:
-
-```json title=".vscode/extensions.json"
-{
-    "recommendations": [
-        "editorconfig.editorconfig",
-        "github.vscode-pull-request-github",
-        "ms-python.black-formatter",
-        "ms-python.flake8",
-        "ms-python.isort",
-        "ms-python.python",
-        "ms-python.vscode-pylance",
-        "redhat.vscode-yaml",
-        "tamasfe.even-better-toml",
-        "eamodio.gitlens",
-        "foxundermoon.shell-format",
-        "timonwong.shellcheck",
-        "esbenp.prettier-vscode",
-        "davidanson.vscode-markdownlint",
-        "yzhang.markdown-all-in-one",
-        "bierner.github-markdown-preview",
-        "ms-azuretools.vscode-docker",
-        "mads-hartmann.bash-ide-vscode"
-    ]
-}
-```
-
-## settings.json
-
-With bellow settings your files already get formated when you save them (only
-your modifications if available), which will help you to not run into trouble
-with the pre-commit hooks. If the hooks fail, they will prevent you from
-commiting, but most hooks directly add a fixed version, so that you just need to
-stage and commit them:
-
-```json title=".vscode/settings.json"
-{
-    "[json]": {
-        "editor.defaultFormatter": "esbenp.prettier-vscode",
-        "editor.quickSuggestions": {
-            "comments": false,
-            "strings": true,
-            "other": true
-        },
-        "editor.suggest.insertMode": "replace",
-        "gitlens.codeLens.scopes": ["document"]
-    },
-    "[jsonc]": {
-        "editor.defaultFormatter": "esbenp.prettier-vscode",
-        "editor.formatOnSave": true,
-        "editor.formatOnSaveMode": "modificationsIfAvailable"
-    },
-    "[python]": {
-        "editor.defaultFormatter": "ms-python.black-formatter",
-        "editor.formatOnSave": true,
-        "editor.formatOnSaveMode": "file"
-    },
-    "[toml]": {
-        "editor.defaultFormatter": "tamasfe.even-better-toml",
-        "editor.formatOnSave": true,
-        "editor.formatOnSaveMode": "modificationsIfAvailable"
-    },
-    "[yaml]": {
-        "editor.defaultFormatter": "esbenp.prettier-vscode",
-        "editor.formatOnSave": true,
-        "editor.formatOnSaveMode": "modificationsIfAvailable"
-    },
-    "[markdown]": {
-        "editor.defaultFormatter": "esbenp.prettier-vscode",
-        "editor.rulers": [80],
-        "editor.unicodeHighlight.ambiguousCharacters": false,
-        "editor.unicodeHighlight.invisibleCharacters": false,
-        "diffEditor.ignoreTrimWhitespace": false,
-        "editor.wordWrap": "on",
-        "editor.quickSuggestions": {
-            "comments": "off",
-            "strings": "off",
-            "other": "off"
-        },
-        "editor.formatOnSave": true,
-        "editor.formatOnSaveMode": "modificationsIfAvailable"
-    },
-    "[shellscript]": {
-        "editor.defaultFormatter": "foxundermoon.shell-format"
-    },
-    "[ignore]": {
-        "editor.defaultFormatter": "foxundermoon.shell-format"
-    },
-    "editor.rulers": [88],
-    "evenBetterToml.formatter.alignEntries": false,
-    "evenBetterToml.formatter.allowedBlankLines": 1,
-    "evenBetterToml.formatter.arrayAutoExpand": true,
-    "evenBetterToml.formatter.arrayTrailingComma": true,
-    "evenBetterToml.formatter.arrayAutoCollapse": true,
-    "evenBetterToml.formatter.columnWidth": 88,
-    "evenBetterToml.formatter.compactArrays": true,
-    "evenBetterToml.formatter.compactInlineTables": true,
-    "evenBetterToml.formatter.indentEntries": false,
-    "evenBetterToml.formatter.inlineTableExpand": true,
-    "evenBetterToml.formatter.reorderArrays": true,
-    "evenBetterToml.formatter.reorderKeys": true,
-    "evenBetterToml.formatter.compactEntries": false,
-    "evenBetterToml.schema.enabled": true,
-    "python.analysis.typeCheckingMode": "basic",
-    "python.formatting.provider": "black",
-    "python.languageServer": "Pylance",
-    "python.linting.enabled": true,
-    "python.linting.flake8Enabled": true,
-    "python.testing.unittestEnabled": false,
-    "python.testing.pytestEnabled": true,
-    "python.testing.pytestArgs": [
-        "tests",
-        "--cov=ldm",
-        "--cov-branch",
-        "--cov-report=term:skip-covered"
-    ],
-    "yaml.schemas": {
-        "https://json.schemastore.org/prettierrc.json": "${workspaceFolder}/.prettierrc.yaml"
-    }
-}
-```

docs/help/contributing/010_PULL_REQUEST.md

@@ -1,135 +0,0 @@
----
-title: Pull-Request
----
-
-# :octicons-git-pull-request-16: Pull-Request
-
-## pre-requirements
-
-To follow the steps in this tutorial you will need:
-
--   [GitHub](https://github.com) account
--   [git](https://git-scm.com/downloads) source controll
--   Text / Code Editor (personally I preffer
-    [Visual Studio Code](https://code.visualstudio.com/Download))
--   Terminal:
-    -   If you are on Linux/MacOS you can use bash or zsh
-    -   for Windows Users the commands are written for PowerShell
-
-## Fork Repository
-
-The first step to be done if you want to contribute to InvokeAI, is to fork the
-rpeository.
-
-Since you are already reading this doc, the easiest way to do so is by clicking
-[here](https://github.com/invoke-ai/InvokeAI/fork). You could also open
-[InvokeAI](https://github.com/invoke-ai/InvoekAI) and click on the "Fork" Button
-in the top right.
-
-## Clone your fork
-
-After you forked the Repository, you should clone it to your dev machine:
-
-=== ":fontawesome-brands-linux:Linux / :simple-apple:macOS"
-
-    ``` sh
-    git clone https://github.com/<github username>/InvokeAI \
-    && cd InvokeAI
-    ```
-
-=== ":fontawesome-brands-windows:Windows"
-
-    ``` powershell
-    git clone https://github.com/<github username>/InvokeAI `
-    && cd InvokeAI
-    ```
-
-## Install in Editable Mode
-
-To install InvokeAI in editable mode, (as always) we recommend to create and
-activate a venv first. Afterwards you can install the InvokeAI Package,
-including dev and docs extras in editable mode, follwed by the installation of
-the pre-commit hook:
-
-=== ":fontawesome-brands-linux:Linux / :simple-apple:macOS"
-
-    ``` sh
-    python -m venv .venv \
-      --prompt InvokeAI \
-      --upgrade-deps \
-    && source .venv/bin/activate \
-    && pip install \
-      --upgrade-deps \
-      --use-pep517 \
-      --editable=".[dev,docs]" \
-    && pre-commit install
-    ```
-
-=== ":fontawesome-brands-windows:Windows"
-
-    ``` powershell
-    python -m venv .venv `
-      --prompt InvokeAI `
-      --upgrade-deps `
-    && .venv/scripts/activate.ps1 `
-    && pip install `
-      --upgrade `
-      --use-pep517 `
-      --editable=".[dev,docs]" `
-    && pre-commit install
-    ```
-
-## Create a branch
-
-Make sure you are on main branch, from there create your feature branch:
-
-=== ":fontawesome-brands-linux:Linux / :simple-apple:macOS"
-
-    ``` sh
-    git checkout main \
-    && git pull \
-    && git checkout -B <branch name>
-    ```
-
-=== ":fontawesome-brands-windows:Windows"
-
-    ``` powershell
-    git checkout main `
-    && git pull `
-    && git checkout -B <branch name>
-    ```
-
-## Commit your changes
-
-When you are done with adding / updating content, you need to commit those
-changes to your repository before you can actually open an PR:
-
-```{ .sh .annotate }
-git add <files you have changed> # (1)!
-git commit -m "A commit message which describes your change"
-git push
-```
-
-1. Replace this with a space seperated list of the files you changed, like:
-   `README.md foo.sh bar.json baz`
-
-## Create a Pull Request
-
-After pushing your changes, you are ready to create a Pull Request. just head
-over to your fork on [GitHub](https://github.com), which should already show you
-a message that there have been recent changes on your feature branch and a green
-button which you could use to create the PR.
-
-The default target for your PRs would be the main branch of
-[invoke-ai/InvokeAI](https://github.com/invoke-ai/InvokeAI)
-
-Another way would be to create it in VS-Code or via the GitHub CLI (or even via
-the GitHub CLI in a VS-Code Terminal Window 🤭):
-
-```sh
-gh pr create
-```
-
-The CLI will inform you if there are still unpushed commits on your branch. It
-will also prompt you for things like the the Title and the Body (Description) if
-you did not already pass them as arguments.

docs/help/contributing/020_ISSUES.md

@@ -1,26 +0,0 @@
----
-title: Issues
----
-
-# :octicons-issue-opened-16: Issues
-
-## :fontawesome-solid-bug: Report a bug
-
-If you stumbled over a bug while using InvokeAI, we would apreciate it a lot if
-you
-[open a issue](https://github.com/invoke-ai/InvokeAI/issues/new?assignees=&labels=bug&template=BUG_REPORT.yml&title=%5Bbug%5D%3A+)
-to inform us about the details so that our developers can look into it.
-
-If you also know how to fix the bug, take a look [here](010_PULL_REQUEST.md) to
-find out how to create a Pull Request.
-
-## Request a feature
-
-If you have a idea for a new feature on your mind which you would like to see in
-InvokeAI, there is a
-[feature request](https://github.com/invoke-ai/InvokeAI/issues/new?assignees=&labels=bug&template=BUG_REPORT.yml&title=%5Bbug%5D%3A+)
-available in the issues section of the repository.
-
-If you are just curious which features already got requested you can find the
-overview of open requests
-[here](https://github.com/invoke-ai/InvokeAI/labels/enhancement)

docs/help/contributing/030_DOCS.md

@@ -1,32 +0,0 @@
----
-title: docs
----
-
-# :simple-readthedocs: MkDocs-Material
-
-If you want to contribute to the docs, there is a easy way to verify the results
-of your changes before commiting them.
-
-Just follow the steps in the [Pull-Requests](010_PULL_REQUEST.md) docs, there we
-already
-[create a venv and install the docs extras](010_PULL_REQUEST.md#install-in-editable-mode).
-When installed it's as simple as:
-
-```sh
-mkdocs serve
-```
-
-This will build the docs locally and serve them on your local host, even
-auto-refresh is included, so you can just update a doc, save it and tab to the
-browser, without the needs of restarting the `mkdocs serve`.
-
-More information about the "mkdocs flavored markdown syntax" can be found
-[here](https://squidfunk.github.io/mkdocs-material/reference/).
-
-## :material-microsoft-visual-studio-code:VS-Code
-
-We also provide a
-[launch configuration for VS-Code](../IDE-Settings/vs-code.md#launchjson) which
-includes a `mkdocs serve` entrypoint as well. You also don't have to worry about
-the formatting since this is automated via prettier, but this is of course not
-limited to VS-Code.

docs/help/contributing/090_NODE_TRANSFORMATION.md

@@ -1,76 +0,0 @@
-# Tranformation to nodes
-
-## Current state
-
-```mermaid
-flowchart TD
-    web[WebUI];
-    cli[CLI];
-    web --> |img2img| generate(generate);
-    web --> |txt2img| generate(generate);
-    cli --> |txt2img| generate(generate);
-    cli --> |img2img| generate(generate);
-    generate --> model_manager;
-    generate --> generators;
-    generate --> ti_manager[TI Manager];
-    generate --> etc;
-```
-
-## Transitional Architecture
-
-### first step
-
-```mermaid
-flowchart TD
-    web[WebUI];
-    cli[CLI];
-    web --> |img2img| img2img_node(Img2img node);
-    web --> |txt2img| generate(generate);
-    img2img_node --> model_manager;
-    img2img_node --> generators;
-    cli --> |txt2img| generate;
-    cli --> |img2img| generate;
-    generate --> model_manager;
-    generate --> generators;
-    generate --> ti_manager[TI Manager];
-    generate --> etc;
-```
-
-### second step
-
-```mermaid
-flowchart TD
-    web[WebUI];
-    cli[CLI];
-    web --> |img2img| img2img_node(img2img node);
-    img2img_node --> model_manager;
-    img2img_node --> generators;
-    web --> |txt2img| txt2img_node(txt2img node);
-    cli --> |txt2img| txt2img_node;
-    cli --> |img2img| generate(generate);
-    generate --> model_manager;
-    generate --> generators;
-    generate --> ti_manager[TI Manager];
-    generate --> etc;
-    txt2img_node --> model_manager;
-    txt2img_node --> generators;
-    txt2img_node --> ti_manager[TI Manager];
-```
-
-## Final Architecture
-
-```mermaid
-flowchart TD
-    web[WebUI];
-    cli[CLI];
-    web --> |img2img|img2img_node(img2img node);
-    cli --> |img2img|img2img_node;
-    web --> |txt2img|txt2img_node(txt2img node);
-    cli --> |txt2img|txt2img_node;
-    img2img_node --> model_manager;
-    txt2img_node --> model_manager;
-    img2img_node --> generators;
-    txt2img_node --> generators;
-    img2img_node --> ti_manager[TI Manager];
-    txt2img_node --> ti_manager[TI Manager];
-```

docs/help/contributing/index.md

@@ -1,16 +0,0 @@
----
-title: Contributing
----
-
-# :fontawesome-solid-code-commit: Contributing
-
-There are different ways how you can contribute to
-[InvokeAI](https://github.com/invoke-ai/InvokeAI), like Translations, opening
-Issues for Bugs or ideas how to improve.
-
-This Section of the docs will explain some of the different ways of how you can
-contribute to make it easier for newcommers as well as advanced users :nerd:
-
-If you want to contribute code, but you do not have an exact idea yet, take a
-look at the currently open
-[:fontawesome-solid-bug: Bug Reports](https://github.com/invoke-ai/InvokeAI/issues?q=is%3Aissue+is%3Aopen+label%3Abug)

docs/help/index.md

@@ -1,12 +0,0 @@
-# :material-help:Help
-
-If you are looking for help with the installation of InvokeAI, please take a
-look into the [Installation](../installation/index.md) section of the docs.
-
-Here you will find help to topics like
-
--   how to contribute
--   configuration recommendation for IDEs
-
-If you have an Idea about what's missing and aren't scared from contributing,
-just take a look at [DOCS](./contributing/030_DOCS.md) to find out how to do so.

docs/index.md

@@ -2,8 +2,6 @@
 title: Home
 ---
 
-# :octicons-home-16: Home
-
 <!--
   The Docs you find here (/docs/*) are built and deployed via mkdocs. If you want to run a local version to verify your changes, it's as simple as::
 
@@ -31,36 +29,36 @@ title: Home
 [![github open prs badge]][github open prs link]
 
 [ci checks on dev badge]:
-    https://flat.badgen.net/github/checks/invoke-ai/InvokeAI/development?label=CI%20status%20on%20dev&cache=900&icon=github
+  https://flat.badgen.net/github/checks/invoke-ai/InvokeAI/development?label=CI%20status%20on%20dev&cache=900&icon=github
 [ci checks on dev link]:
-    https://github.com/invoke-ai/InvokeAI/actions?query=branch%3Adevelopment
+  https://github.com/invoke-ai/InvokeAI/actions?query=branch%3Adevelopment
 [ci checks on main badge]:
-    https://flat.badgen.net/github/checks/invoke-ai/InvokeAI/main?label=CI%20status%20on%20main&cache=900&icon=github
+  https://flat.badgen.net/github/checks/invoke-ai/InvokeAI/main?label=CI%20status%20on%20main&cache=900&icon=github
 [ci checks on main link]:
-    https://github.com/invoke-ai/InvokeAI/actions/workflows/test-invoke-conda.yml
+  https://github.com/invoke-ai/InvokeAI/actions/workflows/test-invoke-conda.yml
 [discord badge]: https://flat.badgen.net/discord/members/ZmtBAhwWhy?icon=discord
 [discord link]: https://discord.gg/ZmtBAhwWhy
 [github forks badge]:
-    https://flat.badgen.net/github/forks/invoke-ai/InvokeAI?icon=github
+  https://flat.badgen.net/github/forks/invoke-ai/InvokeAI?icon=github
 [github forks link]:
-    https://useful-forks.github.io/?repo=lstein%2Fstable-diffusion
+  https://useful-forks.github.io/?repo=lstein%2Fstable-diffusion
 [github open issues badge]:
-    https://flat.badgen.net/github/open-issues/invoke-ai/InvokeAI?icon=github
+  https://flat.badgen.net/github/open-issues/invoke-ai/InvokeAI?icon=github
 [github open issues link]:
-    https://github.com/invoke-ai/InvokeAI/issues?q=is%3Aissue+is%3Aopen
+  https://github.com/invoke-ai/InvokeAI/issues?q=is%3Aissue+is%3Aopen
 [github open prs badge]:
-    https://flat.badgen.net/github/open-prs/invoke-ai/InvokeAI?icon=github
+  https://flat.badgen.net/github/open-prs/invoke-ai/InvokeAI?icon=github
 [github open prs link]:
-    https://github.com/invoke-ai/InvokeAI/pulls?q=is%3Apr+is%3Aopen
+  https://github.com/invoke-ai/InvokeAI/pulls?q=is%3Apr+is%3Aopen
 [github stars badge]:
-    https://flat.badgen.net/github/stars/invoke-ai/InvokeAI?icon=github
+  https://flat.badgen.net/github/stars/invoke-ai/InvokeAI?icon=github
 [github stars link]: https://github.com/invoke-ai/InvokeAI/stargazers
 [latest commit to dev badge]:
-    https://flat.badgen.net/github/last-commit/invoke-ai/InvokeAI/development?icon=github&color=yellow&label=last%20dev%20commit&cache=900
+  https://flat.badgen.net/github/last-commit/invoke-ai/InvokeAI/development?icon=github&color=yellow&label=last%20dev%20commit&cache=900
 [latest commit to dev link]:
-    https://github.com/invoke-ai/InvokeAI/commits/development
+  https://github.com/invoke-ai/InvokeAI/commits/development
 [latest release badge]:
-    https://flat.badgen.net/github/release/invoke-ai/InvokeAI/development?icon=github
+  https://flat.badgen.net/github/release/invoke-ai/InvokeAI/development?icon=github
 [latest release link]: https://github.com/invoke-ai/InvokeAI/releases
 
 </div>
@@ -89,24 +87,24 @@ Q&A</a>]
 
 You wil need one of the following:
 
--   :simple-nvidia: An NVIDIA-based graphics card with 4 GB or more VRAM memory.
--   :simple-amd: An AMD-based graphics card with 4 GB or more VRAM memory (Linux
-    only)
--   :fontawesome-brands-apple: An Apple computer with an M1 chip.
+- :simple-nvidia: An NVIDIA-based graphics card with 4 GB or more VRAM memory.
+- :simple-amd: An AMD-based graphics card with 4 GB or more VRAM memory (Linux
+  only)
+- :fontawesome-brands-apple: An Apple computer with an M1 chip.
 
 We do **not recommend** the following video cards due to issues with their
 running in half-precision mode and having insufficient VRAM to render 512x512
 images in full-precision mode:
 
--   NVIDIA 10xx series cards such as the 1080ti
--   GTX 1650 series cards
--   GTX 1660 series cards
+- NVIDIA 10xx series cards such as the 1080ti
+- GTX 1650 series cards
+- GTX 1660 series cards
 
 ### :fontawesome-solid-memory: Memory and Disk
 
--   At least 12 GB Main Memory RAM.
--   At least 18 GB of free disk space for the machine learning model, Python,
-    and all its dependencies.
+- At least 12 GB Main Memory RAM.
+- At least 18 GB of free disk space for the machine learning model, Python, and
+  all its dependencies.
 
 ## :octicons-package-dependencies-24: Installation
 
@@ -115,65 +113,48 @@ either an Nvidia-based card (with CUDA support) or an AMD card (using the ROCm
 driver).
 
 ### [Installation Getting Started Guide](installation)
-
 #### [Automated Installer](installation/010_INSTALL_AUTOMATED.md)
-
 This method is recommended for 1st time users
-
 #### [Manual Installation](installation/020_INSTALL_MANUAL.md)
-
 This method is recommended for experienced users and developers
-
 #### [Docker Installation](installation/040_INSTALL_DOCKER.md)
-
 This method is recommended for those familiar with running Docker containers
-
 ### Other Installation Guides
-
--   [PyPatchMatch](installation/060_INSTALL_PATCHMATCH.md)
--   [XFormers](installation/070_INSTALL_XFORMERS.md)
--   [CUDA and ROCm Drivers](installation/030_INSTALL_CUDA_AND_ROCM.md)
--   [Installing New Models](installation/050_INSTALLING_MODELS.md)
+  - [PyPatchMatch](installation/060_INSTALL_PATCHMATCH.md)
+  - [XFormers](installation/070_INSTALL_XFORMERS.md)
+  - [CUDA and ROCm Drivers](installation/030_INSTALL_CUDA_AND_ROCM.md)
+  - [Installing New Models](installation/050_INSTALLING_MODELS.md)
 
 ## :octicons-gift-24: InvokeAI Features
 
 ### The InvokeAI Web Interface
-
--   [WebUI overview](features/WEB.md)
--   [WebUI hotkey reference guide](features/WEBUIHOTKEYS.md)
--   [WebUI Unified Canvas for Img2Img, inpainting and outpainting](features/UNIFIED_CANVAS.md)
+- [WebUI overview](features/WEB.md)
+- [WebUI hotkey reference guide](features/WEBUIHOTKEYS.md)
+- [WebUI Unified Canvas for Img2Img, inpainting and outpainting](features/UNIFIED_CANVAS.md)
 <!-- separator -->
-
 ### The InvokeAI Command Line Interface
-
--   [Command Line Interace Reference Guide](features/CLI.md)
+- [Command Line Interace Reference Guide](features/CLI.md)
 <!-- separator -->
-
 ### Image Management
-
--   [Image2Image](features/IMG2IMG.md)
--   [Inpainting](features/INPAINTING.md)
--   [Outpainting](features/OUTPAINTING.md)
--   [Adding custom styles and subjects](features/CONCEPTS.md)
--   [Upscaling and Face Reconstruction](features/POSTPROCESS.md)
--   [Embiggen upscaling](features/EMBIGGEN.md)
--   [Other Features](features/OTHER.md)
+- [Image2Image](features/IMG2IMG.md)
+- [Inpainting](features/INPAINTING.md)
+- [Outpainting](features/OUTPAINTING.md)
+- [Adding custom styles and subjects](features/CONCEPTS.md)
+- [Upscaling and Face Reconstruction](features/POSTPROCESS.md)
+- [Embiggen upscaling](features/EMBIGGEN.md)
+- [Other Features](features/OTHER.md)
 
 <!-- separator -->
-
 ### Model Management
-
--   [Installing](installation/050_INSTALLING_MODELS.md)
--   [Model Merging](features/MODEL_MERGING.md)
--   [Style/Subject Concepts and Embeddings](features/CONCEPTS.md)
--   [Textual Inversion](features/TEXTUAL_INVERSION.md)
--   [Not Safe for Work (NSFW) Checker](features/NSFW.md)
+- [Installing](installation/050_INSTALLING_MODELS.md)
+- [Model Merging](features/MODEL_MERGING.md)
+- [Style/Subject Concepts and Embeddings](features/CONCEPTS.md)
+- [Textual Inversion](features/TEXTUAL_INVERSION.md)
+- [Not Safe for Work (NSFW) Checker](features/NSFW.md)
 <!-- seperator -->
-
 ### Prompt Engineering
-
--   [Prompt Syntax](features/PROMPTS.md)
--   [Generating Variations](features/VARIATIONS.md)
+- [Prompt Syntax](features/PROMPTS.md)
+- [Generating Variations](features/VARIATIONS.md)
 
 ## :octicons-log-16: Latest Changes
 
@@ -181,188 +162,84 @@ This method is recommended for those familiar with running Docker containers
 
 #### Migration to Stable Diffusion `diffusers` models
 
-Previous versions of InvokeAI supported the original model file format
-introduced with Stable Diffusion 1.4. In the original format, known variously as
-"checkpoint", or "legacy" format, there is a single large weights file ending
-with `.ckpt` or `.safetensors`. Though this format has served the community
-well, it has a number of disadvantages, including file size, slow loading times,
-and a variety of non-standard variants that require special-case code to handle.
-In addition, because checkpoint files are actually a bundle of multiple machine
-learning sub-models, it is hard to swap different sub-models in and out, or to
-share common sub-models. A new format, introduced by the StabilityAI company in
-collaboration with HuggingFace, is called `diffusers` and consists of a
-directory of individual models. The most immediate benefit of `diffusers` is
-that they load from disk very quickly. A longer term benefit is that in the near
-future `diffusers` models will be able to share common sub-models, dramatically
-reducing disk space when you have multiple fine-tune models derived from the
-same base.
+Previous versions of InvokeAI supported the original model file format introduced with Stable Diffusion 1.4. In the original format, known variously as "checkpoint", or "legacy" format, there is a single large weights file ending with `.ckpt` or `.safetensors`. Though this format has served the community well, it has a number of disadvantages, including file size, slow loading times, and a variety of non-standard variants that require special-case code to handle. In addition, because checkpoint files are actually a bundle of multiple machine learning sub-models, it is hard to swap different sub-models in and out, or to share common sub-models. A new format, introduced by the StabilityAI company in collaboration with HuggingFace, is called `diffusers` and consists of a directory of individual models. The most immediate benefit of `diffusers` is that they load from disk very quickly. A longer term benefit is that in the near future `diffusers` models will be able to share common sub-models, dramatically reducing disk space when you have multiple fine-tune models derived from the same base.
 
-When you perform a new install of version 2.3.0, you will be offered the option
-to install the `diffusers` versions of a number of popular SD models, including
-Stable Diffusion versions 1.5 and 2.1 (including the 768x768 pixel version of
-2.1). These will act and work just like the checkpoint versions. Do not be
-concerned if you already have a lot of ".ckpt" or ".safetensors" models on disk!
-InvokeAI 2.3.0 can still load these and generate images from them without any
-extra intervention on your part.
+When you perform a new install of version 2.3.0, you will be offered the option to install the `diffusers` versions of a number of popular SD models, including Stable Diffusion versions 1.5 and 2.1 (including the 768x768 pixel version of 2.1). These will act and work just like the checkpoint versions. Do not be concerned if you already have a lot of ".ckpt" or ".safetensors" models on disk! InvokeAI 2.3.0 can still load these and generate images from them without any extra intervention on your part.
 
-To take advantage of the optimized loading times of `diffusers` models, InvokeAI
-offers options to convert legacy checkpoint models into optimized `diffusers`
-models. If you use the `invokeai` command line interface, the relevant commands
-are:
+To take advantage of the optimized loading times of `diffusers` models, InvokeAI offers options to convert legacy checkpoint models into optimized `diffusers` models. If you use the `invokeai` command line interface, the relevant commands are:
 
--   `!convert_model` -- Take the path to a local checkpoint file or a URL that
-    is pointing to one, convert it into a `diffusers` model, and import it into
-    InvokeAI's models registry file.
--   `!optimize_model` -- If you already have a checkpoint model in your InvokeAI
-    models file, this command will accept its short name and convert it into a
-    like-named `diffusers` model, optionally deleting the original checkpoint
-    file.
--   `!import_model` -- Take the local path of either a checkpoint file or a
-    `diffusers` model directory and import it into InvokeAI's registry file. You
-    may also provide the ID of any diffusers model that has been published on
-    the
-    [HuggingFace models repository](https://huggingface.co/models?pipeline_tag=text-to-image&sort=downloads)
-    and it will be downloaded and installed automatically.
+* `!convert_model` -- Take the path to a local checkpoint file or a URL that is pointing to one, convert it into a `diffusers` model, and import it into InvokeAI's models registry file.
+* `!optimize_model` -- If you already have a checkpoint model in your InvokeAI models file, this command will accept its short name and convert it into a like-named `diffusers` model, optionally deleting the original checkpoint file.
+* `!import_model` -- Take the local path of either a checkpoint file or a `diffusers` model directory and import it into InvokeAI's registry file. You may also provide the ID of any diffusers model that has been published on the [HuggingFace models repository](https://huggingface.co/models?pipeline_tag=text-to-image&sort=downloads) and it will be downloaded and installed automatically.
 
 The WebGUI offers similar functionality for model management.
 
-For advanced users, new command-line options provide additional functionality.
-Launching `invokeai` with the argument `--autoconvert <path to directory>` takes
-the path to a directory of checkpoint files, automatically converts them into
-`diffusers` models and imports them. Each time the script is launched, the
-directory will be scanned for new checkpoint files to be loaded. Alternatively,
-the `--ckpt_convert` argument will cause any checkpoint or safetensors model
-that is already registered with InvokeAI to be converted into a `diffusers`
-model on the fly, allowing you to take advantage of future diffusers-only
-features without explicitly converting the model and saving it to disk.
+For advanced users, new command-line options provide additional functionality. Launching `invokeai` with the argument `--autoconvert <path to directory>` takes the path to a directory of checkpoint files, automatically converts them into `diffusers` models and imports them. Each time the script is launched, the directory will be scanned for new checkpoint files to be loaded. Alternatively, the `--ckpt_convert` argument will cause any checkpoint or safetensors model that is already registered with InvokeAI to be converted into a `diffusers` model on the fly, allowing you to take advantage of future diffusers-only features without explicitly converting the model and saving it to disk.
 
-Please see
-[INSTALLING MODELS](https://invoke-ai.github.io/InvokeAI/installation/050_INSTALLING_MODELS/)
-for more information on model management in both the command-line and Web
-interfaces.
+Please see [INSTALLING MODELS](https://invoke-ai.github.io/InvokeAI/installation/050_INSTALLING_MODELS/) for more information on model management in both the command-line and Web interfaces.
 
 #### Support for the `XFormers` Memory-Efficient Crossattention Package
 
-On CUDA (Nvidia) systems, version 2.3.0 supports the `XFormers` library. Once
-installed, the`xformers` package dramatically reduces the memory footprint of
-loaded Stable Diffusion models files and modestly increases image generation
-speed. `xformers` will be installed and activated automatically if you specify a
-CUDA system at install time.
+On CUDA (Nvidia) systems, version 2.3.0 supports the `XFormers` library. Once installed,  the`xformers` package dramatically reduces the memory footprint of loaded Stable Diffusion models files and modestly increases image generation speed. `xformers` will be installed and activated automatically if you specify a CUDA system at install time.
 
-The caveat with using `xformers` is that it introduces slightly
-non-deterministic behavior, and images generated using the same seed and other
-settings will be subtly different between invocations. Generally the changes are
-unnoticeable unless you rapidly shift back and forth between images, but to
-disable `xformers` and restore fully deterministic behavior, you may launch
-InvokeAI using the `--no-xformers` option. This is most conveniently done by
-opening the file `invokeai/invokeai.init` with a text editor, and adding the
-line `--no-xformers` at the bottom.
+The caveat with using `xformers` is that it introduces slightly non-deterministic behavior, and images generated using the same seed and other settings will be subtly different between invocations. Generally the changes are unnoticeable unless you rapidly shift back and forth between images, but to disable `xformers` and restore fully deterministic behavior, you may launch InvokeAI using the `--no-xformers` option. This is most conveniently done by opening the file `invokeai/invokeai.init` with a text editor, and adding the line `--no-xformers` at the bottom.
 
 #### A Negative Prompt Box in the WebUI
 
-There is now a separate text input box for negative prompts in the WebUI. This
-is convenient for stashing frequently-used negative prompts ("mangled limbs, bad
-anatomy"). The `[negative prompt]` syntax continues to work in the main prompt
-box as well.
+There is now a separate text input box for negative prompts in the WebUI. This is convenient for stashing frequently-used negative prompts ("mangled limbs, bad anatomy"). The `[negative prompt]` syntax continues to work in the main prompt box as well.
 
-To see exactly how your prompts are being parsed, launch `invokeai` with the
-`--log_tokenization` option. The console window will then display the
-tokenization process for both positive and negative prompts.
+To see exactly how your prompts are being parsed, launch `invokeai` with the `--log_tokenization` option. The console window will then display the tokenization process for both positive and negative prompts.
 
 #### Model Merging
 
-Version 2.3.0 offers an intuitive user interface for merging up to three Stable
-Diffusion models using an intuitive user interface. Model merging allows you to
-mix the behavior of models to achieve very interesting effects. To use this,
-each of the models must already be imported into InvokeAI and saved in
-`diffusers` format, then launch the merger using a new menu item in the InvokeAI
-launcher script (`invoke.sh`, `invoke.bat`) or directly from the command line
-with `invokeai-merge --gui`. You will be prompted to select the models to merge,
-the proportions in which to mix them, and the mixing algorithm. The script will
-create a new merged `diffusers` model and import it into InvokeAI for your use.
+Version 2.3.0 offers an intuitive user interface for merging up to three Stable Diffusion models using an intuitive user interface. Model merging allows you to mix the behavior of models to achieve very interesting effects. To use this, each of the models must already be imported into InvokeAI and saved in `diffusers` format, then launch the merger using a new menu item in the InvokeAI launcher script (`invoke.sh`, `invoke.bat`) or directly from the command line with `invokeai-merge --gui`. You will be prompted to select the models to merge, the proportions in which to mix them, and the mixing algorithm. The script will create a new merged `diffusers` model and import it into InvokeAI for your use.
 
-See
-[MODEL MERGING](https://invoke-ai.github.io/InvokeAI/features/MODEL_MERGING/)
-for more details.
+See [MODEL MERGING](https://invoke-ai.github.io/InvokeAI/features/MODEL_MERGING/) for more details.
 
 #### Textual Inversion Training
 
-Textual Inversion (TI) is a technique for training a Stable Diffusion model to
-emit a particular subject or style when triggered by a keyword phrase. You can
-perform TI training by placing a small number of images of the subject or style
-in a directory, and choosing a distinctive trigger phrase, such as
-"pointillist-style". After successful training, The subject or style will be
-activated by including `<pointillist-style>` in your prompt.
+Textual Inversion (TI) is a technique for training a Stable Diffusion model to emit a particular subject or style when triggered by a keyword phrase. You can perform TI training by placing a small number of images of the subject or style in a directory, and choosing a distinctive trigger phrase, such as "pointillist-style". After successful training, The subject or style will be activated by including `<pointillist-style>` in your prompt.
 
-Previous versions of InvokeAI were able to perform TI, but it required using a
-command-line script with dozens of obscure command-line arguments. Version 2.3.0
-features an intuitive TI frontend that will build a TI model on top of any
-`diffusers` model. To access training you can launch from a new item in the
-launcher script or from the command line using `invokeai-ti --gui`.
+Previous versions of InvokeAI were able to perform TI, but it required using a command-line script with dozens of obscure command-line arguments. Version 2.3.0 features an intuitive TI frontend that will build a TI model on top of any `diffusers` model. To access training you can launch from a new item in the launcher script or from the command line using `invokeai-ti --gui`.
 
-See
-[TEXTUAL INVERSION](https://invoke-ai.github.io/InvokeAI/features/TEXTUAL_INVERSION/)
-for further details.
+See [TEXTUAL INVERSION](https://invoke-ai.github.io/InvokeAI/features/TEXTUAL_INVERSION/) for further details.
 
 #### A New Installer Experience
 
-The InvokeAI installer has been upgraded in order to provide a smoother and
-hopefully more glitch-free experience. In addition, InvokeAI is now packaged as
-a PyPi project, allowing developers and power-users to install InvokeAI with the
-command `pip install InvokeAI  --use-pep517`. Please see
-[Installation](#installation) for details.
+The InvokeAI installer has been upgraded in order to provide a smoother and hopefully more glitch-free experience. In addition, InvokeAI is now packaged as a PyPi project, allowing developers and power-users to install InvokeAI with the command `pip install InvokeAI  --use-pep517`. Please see [Installation](#installation) for details.
 
-Developers should be aware that the `pip` installation procedure has been
-simplified and that the `conda` method is no longer supported at all.
-Accordingly, the `environments_and_requirements` directory has been deleted from
-the repository.
+Developers should be aware that the `pip` installation procedure has been simplified and that the `conda` method is no longer supported at all. Accordingly, the `environments_and_requirements` directory has been deleted from the repository.
 
 #### Command-line name changes
 
-All of InvokeAI's functionality, including the WebUI, command-line interface,
-textual inversion training and model merging, can all be accessed from the
-`invoke.sh` and `invoke.bat` launcher scripts. The menu of options has been
-expanded to add the new functionality. For the convenience of developers and
-power users, we have normalized the names of the InvokeAI command-line scripts:
+All of InvokeAI's functionality, including the WebUI, command-line interface, textual inversion training and model merging, can all be accessed from the `invoke.sh` and `invoke.bat` launcher scripts. The menu of options has been expanded to add the new functionality. For the convenience of developers and power users, we have normalized the names of the InvokeAI command-line scripts:
 
--   `invokeai` -- Command-line client
--   `invokeai --web` -- Web GUI
--   `invokeai-merge --gui` -- Model merging script with graphical front end
--   `invokeai-ti --gui` -- Textual inversion script with graphical front end
--   `invokeai-configure` -- Configuration tool for initializing the `invokeai`
-    directory and selecting popular starter models.
+* `invokeai` -- Command-line client
+* `invokeai --web` -- Web GUI
+* `invokeai-merge --gui` -- Model merging script with graphical front end
+* `invokeai-ti --gui` -- Textual inversion script with graphical front end
+* `invokeai-configure` -- Configuration tool for initializing the `invokeai` directory and selecting popular starter models.
 
-For backward compatibility, the old command names are also recognized, including
-`invoke.py` and `configure-invokeai.py`. However, these are deprecated and will
-eventually be removed.
+For backward compatibility, the old command names are also recognized, including `invoke.py` and `configure-invokeai.py`. However, these are deprecated and will eventually be removed.
 
-Developers should be aware that the locations of the script's source code has
-been moved. The new locations are:
+Developers should be aware that the locations of the script's source code has been moved. The new locations are:
+* `invokeai` =>  `ldm/invoke/CLI.py`
+* `invokeai-configure` => `ldm/invoke/config/configure_invokeai.py`
+* `invokeai-ti`=> `ldm/invoke/training/textual_inversion.py`
+* `invokeai-merge` => `ldm/invoke/merge_diffusers`
 
--   `invokeai` => `ldm/invoke/CLI.py`
--   `invokeai-configure` => `ldm/invoke/config/configure_invokeai.py`
--   `invokeai-ti`=> `ldm/invoke/training/textual_inversion.py`
--   `invokeai-merge` => `ldm/invoke/merge_diffusers`
+Developers are strongly encouraged to perform an "editable" install of InvokeAI using `pip install -e .  --use-pep517` in the Git repository, and then to call the scripts using their 2.3.0 names, rather than executing the scripts directly. Developers should also be aware that the several important data files have been relocated into a new directory named `invokeai`. This includes the WebGUI's `frontend` and `backend` directories, and the `INITIAL_MODELS.yaml` files used by the installer to select starter models. Eventually all InvokeAI modules will be in subdirectories of `invokeai`.
 
-Developers are strongly encouraged to perform an "editable" install of InvokeAI
-using `pip install -e .  --use-pep517` in the Git repository, and then to call
-the scripts using their 2.3.0 names, rather than executing the scripts directly.
-Developers should also be aware that the several important data files have been
-relocated into a new directory named `invokeai`. This includes the WebGUI's
-`frontend` and `backend` directories, and the `INITIAL_MODELS.yaml` files used
-by the installer to select starter models. Eventually all InvokeAI modules will
-be in subdirectories of `invokeai`.
-
-Please see
-[2.3.0 Release Notes](https://github.com/invoke-ai/InvokeAI/releases/tag/v2.3.0)
-for further details. For older changelogs, please visit the
+Please see [2.3.0 Release Notes](https://github.com/invoke-ai/InvokeAI/releases/tag/v2.3.0) for further details.
+For older changelogs, please visit the
 **[CHANGELOG](CHANGELOG/#v223-2-december-2022)**.
 
 ## :material-target: Troubleshooting
 
-Please check out our
-**[:material-frequently-asked-questions: Troubleshooting Guide](installation/010_INSTALL_AUTOMATED.md#troubleshooting)**
-to get solutions for common installation problems and other issues.
+Please check out our **[:material-frequently-asked-questions:
+Troubleshooting
+Guide](installation/010_INSTALL_AUTOMATED.md#troubleshooting)** to
+get solutions for common installation problems and other issues.
 
 ## :octicons-repo-push-24: Contributing
 
@@ -388,8 +265,8 @@ thank them for their time, hard work and effort.
 For support, please use this repository's GitHub Issues tracking service. Feel
 free to send me an email if you use and like the script.
 
-Original portions of the software are Copyright (c) 2022-23 by
-[The InvokeAI Team](https://github.com/invoke-ai).
+Original portions of the software are Copyright (c) 2022-23
+by [The InvokeAI Team](https://github.com/invoke-ai).
 
 ## :octicons-book-24: Further Reading
 

docs/installation/010_INSTALL_AUTOMATED.md

@@ -417,7 +417,7 @@ Then type the following commands:
 
 === "AMD System"
     ```bash
-    pip install torch torchvision --force-reinstall --extra-index-url https://download.pytorch.org/whl/rocm5.4.2
+    pip install torch torchvision --force-reinstall --extra-index-url https://download.pytorch.org/whl/rocm5.2
     ```
 
 ### Corrupted configuration file

docs/installation/020_INSTALL_MANUAL.md

@@ -148,7 +148,7 @@ manager, please follow these steps:
     === "CUDA (NVidia)"
 
         ```bash
-        pip install InvokeAI[xformers] --use-pep517 --extra-index-url https://download.pytorch.org/whl/cu117
+        pip install "InvokeAI[xformers]" --use-pep517 --extra-index-url https://download.pytorch.org/whl/cu117
         ```
 
     === "ROCm (AMD)"

docs/installation/030_INSTALL_CUDA_AND_ROCM.md

@@ -110,7 +110,7 @@ recipes are available
 
 When installing torch and torchvision manually with `pip`, remember to provide
 the argument `--extra-index-url
-https://download.pytorch.org/whl/rocm5.4.2` as described in the [Manual
+https://download.pytorch.org/whl/rocm5.2` as described in the [Manual
 Installation Guide](020_INSTALL_MANUAL.md).
 
 This will be done automatically for you if you use the installer

docs/installation/050_INSTALLING_MODELS.md

@@ -211,26 +211,6 @@ description for the model, whether to make this the default model that
 is loaded at InvokeAI startup time, and whether to replace its
 VAE. Generally the answer to the latter question is "no".
 
-### Specifying a configuration file for legacy checkpoints
-
-Some checkpoint files come with instructions to use a specific .yaml
-configuration file. For InvokeAI load this file correctly, please put
-the config file in the same directory as the corresponding `.ckpt` or
-`.safetensors` file and make sure the file has the same basename as
-the weights file. Here is an example:
-
-```bash
-wonderful-model-v2.ckpt
-wonderful-model-v2.yaml
-```
-
-Similarly, to use a custom VAE, name the VAE like this:
-
-```bash
-wonderful-model-v2.vae.pt
-```
-
-
 ### Converting legacy models into `diffusers`
 
 The CLI `!convert_model` will convert a `.safetensors` or `.ckpt`

docs/requirements-mkdocs.txt

@@ -0,0 +1,5 @@
+mkdocs
+mkdocs-material>=8, <9
+mkdocs-git-revision-date-localized-plugin
+mkdocs-redirects==1.2.0
+

installer/create_installer.sh

@@ -11,10 +11,10 @@ if [[ -v "VIRTUAL_ENV" ]]; then
     exit -1
 fi
 
-VERSION=$(cd ..; python -c "from ldm.invoke import __version__ as version; print(version)")
+VERSION=$(cd ..; python -c "from invokeai.version import __version__ as version; print(version)")
 PATCH=""
 VERSION="v${VERSION}${PATCH}"
-LATEST_TAG="v2.3-latest"
+LATEST_TAG="v3.0-latest"
 
 echo Building installer for version $VERSION
 echo "Be certain that you're in the 'installer' directory before continuing."

installer/lib/installer.py

@@ -241,18 +241,14 @@ class InvokeAiInstance:
 
         from plumbum import FG, local
 
-        # Note that we're installing pinned versions of torch and
-        # torchvision here, which may not correspond to what is
-        # in pyproject.toml. This is a hack to prevent torch 2.0 from
-        # being installed and immediately uninstalled and replaced with 1.13
         pip = local[self.pip]
 
         (
             pip[
                 "install",
                 "--require-virtualenv",
-                "torch~=1.13.1",
-                "torchvision>=0.14.1",
+                "torch",
+                "torchvision",
                 "--force-reinstall",
                 "--find-links" if find_links is not None else None,
                 find_links,
@@ -295,7 +291,7 @@ class InvokeAiInstance:
                     src = Path(__file__).parents[1].expanduser().resolve()
                     # if the above directory contains one of these files, we'll do a source install
                     next(src.glob("pyproject.toml"))
-                    next(src.glob("ldm"))
+                    next(src.glob("invokeai"))
                 except StopIteration:
                     print("Unable to find a wheel or perform a source install. Giving up.")
 
@@ -346,14 +342,14 @@ class InvokeAiInstance:
 
         introduction()
 
-        from ldm.invoke.config import invokeai_configure
+        from invokeai.frontend.install import invokeai_configure
 
         # NOTE: currently the config script does its own arg parsing! this means the command-line switches
         # from the installer will also automatically propagate down to the config script.
         # this may change in the future with config refactoring!
         succeeded = False
         try:
-            invokeai_configure.main()
+            invokeai_configure()
             succeeded = True
         except requests.exceptions.ConnectionError as e:
             print(f'\nA network error was encountered during configuration and download: {str(e)}')
@@ -383,9 +379,6 @@ class InvokeAiInstance:
             shutil.copy(src, dest)
             os.chmod(dest, 0o0755)
 
-        if OS == "Linux":
-            shutil.copy(Path(__file__).parent / '..' / "templates" / "dialogrc", self.runtime / '.dialogrc')
-
     def update(self):
         pass
 

installer/templates/dialogrc

@@ -1,27 +0,0 @@
-# Screen 
-use_shadow = OFF
-use_colors = ON
-screen_color = (BLACK, BLACK, ON)
-
-# Box 
-dialog_color = (YELLOW, BLACK , ON) 
-title_color = (YELLOW, BLACK, ON) 
-border_color = (YELLOW, BLACK, OFF) 
-border2_color = (YELLOW, BLACK, OFF) 
-
-# Button 
-button_active_color = (RED, BLACK, OFF) 
-button_inactive_color = (YELLOW, BLACK, OFF) 
-button_label_active_color = (YELLOW,BLACK,ON) 
-button_label_inactive_color = (YELLOW,BLACK,ON)
- 
-# Menu box 
-menubox_color = (BLACK, BLACK, ON) 
-menubox_border_color = (YELLOW, BLACK, OFF) 
-menubox_border2_color = (YELLOW, BLACK, OFF) 
-
-# Menu window 
-item_color = (YELLOW, BLACK, OFF) 
-item_selected_color = (BLACK, YELLOW, OFF) 
-tag_key_color = (YELLOW, BLACK, OFF) 
-tag_key_selected_color = (BLACK, YELLOW, OFF)

installer/templates/invoke.sh.in

@@ -1,10 +1,5 @@
 #!/bin/bash
 
-# MIT License
-
-# Coauthored by Lincoln Stein, Eugene Brodsky and Joshua Kimsey
-# Copyright 2023, The InvokeAI Development Team
-
 ####
 # This launch script assumes that:
 # 1. it is located in the runtime directory,
@@ -16,168 +11,85 @@
 
 set -eu
 
-# Ensure we're in the correct folder in case user's CWD is somewhere else
+# ensure we're in the correct folder in case user's CWD is somewhere else
 scriptdir=$(dirname "$0")
 cd "$scriptdir"
 
 . .venv/bin/activate
 
 export INVOKEAI_ROOT="$scriptdir"
-PARAMS=$@
 
-# Check to see if dialog is installed (it seems to be fairly standard, but good to check regardless) and if the user has passed the --no-tui argument to disable the dialog TUI
-tui=true
-if command -v dialog &>/dev/null; then
-    # This must use $@ to properly loop through the arguments passed by the user
-    for arg in "$@"; do
-        if [ "$arg" == "--no-tui" ]; then
-            tui=false
-            # Remove the --no-tui argument to avoid errors later on when passing arguments to InvokeAI
-            PARAMS=$(echo "$PARAMS" | sed 's/--no-tui//')
-            break
-        fi
-    done
-else
-    tui=false
-fi
-
-# Set required env var for torch on mac MPS
+# set required env var for torch on mac MPS
 if [ "$(uname -s)" == "Darwin" ]; then
     export PYTORCH_ENABLE_MPS_FALLBACK=1
 fi
 
-# Primary function for the case statement to determine user input
-do_choice() {
-    case $1 in
-    1)
-        clear
-        printf "Generate images with a browser-based interface\n"
-        invokeai --web $PARAMS
-        ;;
-    2)
-        clear
-        printf "Generate images using a command-line interface\n"
-        invokeai $PARAMS
-        ;;
-    3)
-        clear
-        printf "Textual inversion training\n"
-        invokeai-ti --gui $PARAMS
-        ;;
-    4)
-        clear
-        printf "Merge models (diffusers type only)\n"
-        invokeai-merge --gui $PARAMS
-        ;;
-    5)
-        clear
-        printf "Download and install models\n"
-        invokeai-model-install --root ${INVOKEAI_ROOT}
-        ;;
-    6)
-        clear
-        printf "Change InvokeAI startup options\n"
-        invokeai-configure --root ${INVOKEAI_ROOT} --skip-sd-weights --skip-support-models
-        ;;
-    7)
-        clear
-        printf "Re-run the configure script to fix a broken install\n"
-        invokeai-configure --root ${INVOKEAI_ROOT} --yes --default_only
-        ;;
-    8)
-        clear
-        printf "Open the developer console\n"
-        file_name=$(basename "${BASH_SOURCE[0]}")
-        bash --init-file "$file_name"
-        ;;
-    9)
-        clear
-        printf "Update InvokeAI\n"
-        invokeai-update
-        ;;
-    10)
-        clear
-        printf "Command-line help\n"
-        invokeai --help
-        ;;
-    "HELP 1")
-        clear
-        printf "Command-line help\n"
-        invokeai --help
-        ;;
-    *)
-        clear
-        printf "Exiting...\n"
-        exit
-        ;;
-    esac
-    clear
-}
-
-# Dialog-based TUI for launcing Invoke functions
-do_dialog() {
-    options=(
-        1 "Generate images with a browser-based interface"
-        2 "Generate images using a command-line interface"
-        3 "Textual inversion training"
-        4 "Merge models (diffusers type only)"
-        5 "Download and install models"
-        6 "Change InvokeAI startup options"
-        7 "Re-run the configure script to fix a broken install"
-        8 "Open the developer console"
-        9 "Update InvokeAI")
-
-    choice=$(dialog --clear \
-        --backtitle "\Zb\Zu\Z3InvokeAI" \
-        --colors \
-        --title "What would you like to run?" \
-        --ok-label "Run" \
-        --cancel-label "Exit" \
-        --help-button \
-        --help-label "CLI Help" \
-        --menu "Select an option:" \
-        0 0 0 \
-        "${options[@]}" \
-        2>&1 >/dev/tty) || clear
-    do_choice "$choice"
-    clear
-}
-
-# Command-line interface for launching Invoke functions
-do_line_input() {
-    clear
-    printf " ** For a more attractive experience, please install the 'dialog' utility using your package manager. **\n\n"
-    printf "Do you want to generate images using the\n"
-    printf "1: Browser-based UI\n"
-    printf "2: Command-line interface\n"
-    printf "3: Run textual inversion training\n"
-    printf "4: Merge models (diffusers type only)\n"
-    printf "5: Download and install models\n"
-    printf "6: Change InvokeAI startup options\n"
-    printf "7: Re-run the configure script to fix a broken install\n"
-    printf "8: Open the developer console\n"
-    printf "9: Update InvokeAI\n"
-    printf "10: Command-line help\n"
-    printf "Q: Quit\n\n"
-    read -p "Please enter 1-10, Q: [1] " yn
-    choice=${yn:='1'}
-    do_choice $choice
-    clear
-}
-
-# Main IF statement for launching Invoke with either the TUI or CLI, and for checking if the user is in the developer console
+while true
+do
 if [ "$0" != "bash" ]; then
-    while true; do
-        if $tui; then
-            # .dialogrc must be located in the same directory as the invoke.sh script
-            export DIALOGRC="./.dialogrc"
-            do_dialog
-        else
-            do_line_input
-        fi
-    done
+    echo "Do you want to generate images using the"
+    echo "1. command-line interface"
+    echo "2. browser-based UI"
+    echo "3. run textual inversion training"
+    echo "4. merge models (diffusers type only)"
+    echo "5. download and install models"
+    echo "6. change InvokeAI startup options"
+    echo "7. re-run the configure script to fix a broken install"
+    echo "8. open the developer console"
+    echo "9. update InvokeAI"
+    echo "10. command-line help"
+    echo "Q - Quit"
+    echo ""
+    read -p "Please enter 1-10, Q: [2] " yn
+    choice=${yn:='2'}
+    case $choice in
+        1)
+            echo "Starting the InvokeAI command-line..."
+            invokeai $@
+            ;;
+        2)
+            echo "Starting the InvokeAI browser-based UI..."
+            invokeai --web $@
+            ;;
+        3)
+            echo "Starting Textual Inversion:"
+            invokeai-ti --gui $@
+            ;;
+        4)
+            echo "Merging Models:"
+            invokeai-merge --gui $@
+            ;;
+        5)
+            invokeai-model-install --root ${INVOKEAI_ROOT}
+            ;;
+        6)
+            invokeai-configure --root ${INVOKEAI_ROOT} --skip-sd-weights --skip-support-models
+            ;;
+        7)
+            invokeai-configure --root ${INVOKEAI_ROOT} --yes --default_only
+	    ;;
+	8)
+	    echo "Developer Console:"
+            file_name=$(basename "${BASH_SOURCE[0]}")
+            bash --init-file "$file_name"
+            ;;
+        9)
+	    echo "Update:"
+            invokeai-update
+            ;;
+        10)
+            invokeai --help
+            ;;
+	[qQ])
+            exit 0
+            ;;
+        *)
+            echo "Invalid selection"
+            exit;;
+    esac
 else # in developer console
     python --version
-    printf "Press ^D to exit\n"
+    echo "Press ^D to exit"
     export PS1="(InvokeAI) \u@\h \w> "
 fi
+done

invokeai/README

@@ -1,3 +1,11 @@
-After version 2.3 is released, the ldm/invoke modules will be migrated to this location
-so that we have a proper invokeai distribution. Currently it is only being used for
-data files.
+Organization of the source tree:
+
+app -- Home of nodes invocations and services
+assets -- Images and other data files used by InvokeAI
+backend -- Non-user facing libraries, including the rendering
+	core.
+configs -- Configuration files used at install and run times
+frontend -- User-facing scripts, including the CLI and the WebUI
+version -- Current InvokeAI version string, stored
+	in version/invokeai_version.py
+	

invokeai/app/api/dependencies.py

@@ -0,0 +1,79 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+import os
+from argparse import Namespace
+
+from ...backend import Globals
+from ..services.model_manager_initializer import get_model_manager
+from ..services.restoration_services import RestorationServices
+from ..services.graph import GraphExecutionState
+from ..services.image_storage import DiskImageStorage
+from ..services.invocation_queue import MemoryInvocationQueue
+from ..services.invocation_services import InvocationServices
+from ..services.invoker import Invoker
+from ..services.processor import DefaultInvocationProcessor
+from ..services.sqlite import SqliteItemStorage
+from .events import FastAPIEventService
+
+
+# TODO: is there a better way to achieve this?
+def check_internet() -> bool:
+    """
+    Return true if the internet is reachable.
+    It does this by pinging huggingface.co.
+    """
+    import urllib.request
+
+    host = "http://huggingface.co"
+    try:
+        urllib.request.urlopen(host, timeout=1)
+        return True
+    except:
+        return False
+
+
+class ApiDependencies:
+    """Contains and initializes all dependencies for the API"""
+
+    invoker: Invoker = None
+
+    @staticmethod
+    def initialize(config, event_handler_id: int):
+        Globals.try_patchmatch = config.patchmatch
+        Globals.always_use_cpu = config.always_use_cpu
+        Globals.internet_available = config.internet_available and check_internet()
+        Globals.disable_xformers = not config.xformers
+        Globals.ckpt_convert = config.ckpt_convert
+
+        # TODO: Use a logger
+        print(f">> Internet connectivity is {Globals.internet_available}")
+
+        events = FastAPIEventService(event_handler_id)
+
+        output_folder = os.path.abspath(
+            os.path.join(os.path.dirname(__file__), "../../../../outputs")
+        )
+
+        images = DiskImageStorage(output_folder)
+
+        # TODO: build a file/path manager?
+        db_location = os.path.join(output_folder, "invokeai.db")
+
+        services = InvocationServices(
+            model_manager=get_model_manager(config),
+            events=events,
+            images=images,
+            queue=MemoryInvocationQueue(),
+            graph_execution_manager=SqliteItemStorage[GraphExecutionState](
+                filename=db_location, table_name="graph_executions"
+            ),
+            processor=DefaultInvocationProcessor(),
+            restoration=RestorationServices(config),
+        )
+
+        ApiDependencies.invoker = Invoker(services)
+
+    @staticmethod
+    def shutdown():
+        if ApiDependencies.invoker:
+            ApiDependencies.invoker.stop()

invokeai/app/api/events.py

@@ -0,0 +1,52 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+import asyncio
+import threading
+from queue import Empty, Queue
+from typing import Any
+
+from fastapi_events.dispatcher import dispatch
+
+from ..services.events import EventServiceBase
+
+
+class FastAPIEventService(EventServiceBase):
+    event_handler_id: int
+    __queue: Queue
+    __stop_event: threading.Event
+
+    def __init__(self, event_handler_id: int) -> None:
+        self.event_handler_id = event_handler_id
+        self.__queue = Queue()
+        self.__stop_event = threading.Event()
+        asyncio.create_task(self.__dispatch_from_queue(stop_event=self.__stop_event))
+
+        super().__init__()
+
+    def stop(self, *args, **kwargs):
+        self.__stop_event.set()
+        self.__queue.put(None)
+
+    def dispatch(self, event_name: str, payload: Any) -> None:
+        self.__queue.put(dict(event_name=event_name, payload=payload))
+
+    async def __dispatch_from_queue(self, stop_event: threading.Event):
+        """Get events on from the queue and dispatch them, from the correct thread"""
+        while not stop_event.is_set():
+            try:
+                event = self.__queue.get(block=False)
+                if not event:  # Probably stopping
+                    continue
+
+                dispatch(
+                    event.get("event_name"),
+                    payload=event.get("payload"),
+                    middleware_id=self.event_handler_id,
+                )
+
+            except Empty:
+                await asyncio.sleep(0.001)
+                pass
+
+            except asyncio.CancelledError as e:
+                raise e  # Raise a proper error

invokeai/app/api/routers/images.py

@@ -0,0 +1,56 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+from datetime import datetime, timezone
+
+from fastapi import Path, Request, UploadFile
+from fastapi.responses import FileResponse, Response
+from fastapi.routing import APIRouter
+from PIL import Image
+
+from ...services.image_storage import ImageType
+from ..dependencies import ApiDependencies
+
+images_router = APIRouter(prefix="/v1/images", tags=["images"])
+
+
+@images_router.get("/{image_type}/{image_name}", operation_id="get_image")
+async def get_image(
+    image_type: ImageType = Path(description="The type of image to get"),
+    image_name: str = Path(description="The name of the image to get"),
+):
+    """Gets a result"""
+    # TODO: This is not really secure at all. At least make sure only output results are served
+    filename = ApiDependencies.invoker.services.images.get_path(image_type, image_name)
+    return FileResponse(filename)
+
+
+@images_router.post(
+    "/uploads/",
+    operation_id="upload_image",
+    responses={
+        201: {"description": "The image was uploaded successfully"},
+        404: {"description": "Session not found"},
+    },
+)
+async def upload_image(file: UploadFile, request: Request):
+    if not file.content_type.startswith("image"):
+        return Response(status_code=415)
+
+    contents = await file.read()
+    try:
+        im = Image.open(contents)
+    except:
+        # Error opening the image
+        return Response(status_code=415)
+
+    filename = f"{str(int(datetime.now(timezone.utc).timestamp()))}.png"
+    ApiDependencies.invoker.services.images.save(ImageType.UPLOAD, filename, im)
+
+    return Response(
+        status_code=201,
+        headers={
+            "Location": request.url_for(
+                "get_image", image_type=ImageType.UPLOAD, image_name=filename
+            )
+        },
+    )

invokeai/app/api/routers/sessions.py

@@ -0,0 +1,287 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+from typing import Annotated, List, Optional, Union
+
+from fastapi import Body, Path, Query
+from fastapi.responses import Response
+from fastapi.routing import APIRouter
+from pydantic.fields import Field
+
+from ...invocations import *
+from ...invocations.baseinvocation import BaseInvocation
+from ...services.graph import (
+    Edge,
+    EdgeConnection,
+    Graph,
+    GraphExecutionState,
+    NodeAlreadyExecutedError,
+)
+from ...services.item_storage import PaginatedResults
+from ..dependencies import ApiDependencies
+
+session_router = APIRouter(prefix="/v1/sessions", tags=["sessions"])
+
+
+@session_router.post(
+    "/",
+    operation_id="create_session",
+    responses={
+        200: {"model": GraphExecutionState},
+        400: {"description": "Invalid json"},
+    },
+)
+async def create_session(
+    graph: Optional[Graph] = Body(
+        default=None, description="The graph to initialize the session with"
+    )
+) -> GraphExecutionState:
+    """Creates a new session, optionally initializing it with an invocation graph"""
+    session = ApiDependencies.invoker.create_execution_state(graph)
+    return session
+
+
+@session_router.get(
+    "/",
+    operation_id="list_sessions",
+    responses={200: {"model": PaginatedResults[GraphExecutionState]}},
+)
+async def list_sessions(
+    page: int = Query(default=0, description="The page of results to get"),
+    per_page: int = Query(default=10, description="The number of results per page"),
+    query: str = Query(default="", description="The query string to search for"),
+) -> PaginatedResults[GraphExecutionState]:
+    """Gets a list of sessions, optionally searching"""
+    if filter == "":
+        result = ApiDependencies.invoker.services.graph_execution_manager.list(
+            page, per_page
+        )
+    else:
+        result = ApiDependencies.invoker.services.graph_execution_manager.search(
+            query, page, per_page
+        )
+    return result
+
+
+@session_router.get(
+    "/{session_id}",
+    operation_id="get_session",
+    responses={
+        200: {"model": GraphExecutionState},
+        404: {"description": "Session not found"},
+    },
+)
+async def get_session(
+    session_id: str = Path(description="The id of the session to get"),
+) -> GraphExecutionState:
+    """Gets a session"""
+    session = ApiDependencies.invoker.services.graph_execution_manager.get(session_id)
+    if session is None:
+        return Response(status_code=404)
+    else:
+        return session
+
+
+@session_router.post(
+    "/{session_id}/nodes",
+    operation_id="add_node",
+    responses={
+        200: {"model": str},
+        400: {"description": "Invalid node or link"},
+        404: {"description": "Session not found"},
+    },
+)
+async def add_node(
+    session_id: str = Path(description="The id of the session"),
+    node: Annotated[
+        Union[BaseInvocation.get_invocations()], Field(discriminator="type") # type: ignore
+    ] = Body(description="The node to add"),
+) -> str:
+    """Adds a node to the graph"""
+    session = ApiDependencies.invoker.services.graph_execution_manager.get(session_id)
+    if session is None:
+        return Response(status_code=404)
+
+    try:
+        session.add_node(node)
+        ApiDependencies.invoker.services.graph_execution_manager.set(
+            session
+        )  # TODO: can this be done automatically, or add node through an API?
+        return session.id
+    except NodeAlreadyExecutedError:
+        return Response(status_code=400)
+    except IndexError:
+        return Response(status_code=400)
+
+
+@session_router.put(
+    "/{session_id}/nodes/{node_path}",
+    operation_id="update_node",
+    responses={
+        200: {"model": GraphExecutionState},
+        400: {"description": "Invalid node or link"},
+        404: {"description": "Session not found"},
+    },
+)
+async def update_node(
+    session_id: str = Path(description="The id of the session"),
+    node_path: str = Path(description="The path to the node in the graph"),
+    node: Annotated[
+        Union[BaseInvocation.get_invocations()], Field(discriminator="type") # type: ignore
+    ] = Body(description="The new node"),
+) -> GraphExecutionState:
+    """Updates a node in the graph and removes all linked edges"""
+    session = ApiDependencies.invoker.services.graph_execution_manager.get(session_id)
+    if session is None:
+        return Response(status_code=404)
+
+    try:
+        session.update_node(node_path, node)
+        ApiDependencies.invoker.services.graph_execution_manager.set(
+            session
+        )  # TODO: can this be done automatically, or add node through an API?
+        return session
+    except NodeAlreadyExecutedError:
+        return Response(status_code=400)
+    except IndexError:
+        return Response(status_code=400)
+
+
+@session_router.delete(
+    "/{session_id}/nodes/{node_path}",
+    operation_id="delete_node",
+    responses={
+        200: {"model": GraphExecutionState},
+        400: {"description": "Invalid node or link"},
+        404: {"description": "Session not found"},
+    },
+)
+async def delete_node(
+    session_id: str = Path(description="The id of the session"),
+    node_path: str = Path(description="The path to the node to delete"),
+) -> GraphExecutionState:
+    """Deletes a node in the graph and removes all linked edges"""
+    session = ApiDependencies.invoker.services.graph_execution_manager.get(session_id)
+    if session is None:
+        return Response(status_code=404)
+
+    try:
+        session.delete_node(node_path)
+        ApiDependencies.invoker.services.graph_execution_manager.set(
+            session
+        )  # TODO: can this be done automatically, or add node through an API?
+        return session
+    except NodeAlreadyExecutedError:
+        return Response(status_code=400)
+    except IndexError:
+        return Response(status_code=400)
+
+
+@session_router.post(
+    "/{session_id}/edges",
+    operation_id="add_edge",
+    responses={
+        200: {"model": GraphExecutionState},
+        400: {"description": "Invalid node or link"},
+        404: {"description": "Session not found"},
+    },
+)
+async def add_edge(
+    session_id: str = Path(description="The id of the session"),
+    edge: Edge = Body(description="The edge to add"),
+) -> GraphExecutionState:
+    """Adds an edge to the graph"""
+    session = ApiDependencies.invoker.services.graph_execution_manager.get(session_id)
+    if session is None:
+        return Response(status_code=404)
+
+    try:
+        session.add_edge(edge)
+        ApiDependencies.invoker.services.graph_execution_manager.set(
+            session
+        )  # TODO: can this be done automatically, or add node through an API?
+        return session
+    except NodeAlreadyExecutedError:
+        return Response(status_code=400)
+    except IndexError:
+        return Response(status_code=400)
+
+
+# TODO: the edge being in the path here is really ugly, find a better solution
+@session_router.delete(
+    "/{session_id}/edges/{from_node_id}/{from_field}/{to_node_id}/{to_field}",
+    operation_id="delete_edge",
+    responses={
+        200: {"model": GraphExecutionState},
+        400: {"description": "Invalid node or link"},
+        404: {"description": "Session not found"},
+    },
+)
+async def delete_edge(
+    session_id: str = Path(description="The id of the session"),
+    from_node_id: str = Path(description="The id of the node the edge is coming from"),
+    from_field: str = Path(description="The field of the node the edge is coming from"),
+    to_node_id: str = Path(description="The id of the node the edge is going to"),
+    to_field: str = Path(description="The field of the node the edge is going to"),
+) -> GraphExecutionState:
+    """Deletes an edge from the graph"""
+    session = ApiDependencies.invoker.services.graph_execution_manager.get(session_id)
+    if session is None:
+        return Response(status_code=404)
+
+    try:
+        edge = Edge(
+            source=EdgeConnection(node_id=from_node_id, field=from_field),
+            destination=EdgeConnection(node_id=to_node_id, field=to_field)
+        )
+        session.delete_edge(edge)
+        ApiDependencies.invoker.services.graph_execution_manager.set(
+            session
+        )  # TODO: can this be done automatically, or add node through an API?
+        return session
+    except NodeAlreadyExecutedError:
+        return Response(status_code=400)
+    except IndexError:
+        return Response(status_code=400)
+
+
+@session_router.put(
+    "/{session_id}/invoke",
+    operation_id="invoke_session",
+    responses={
+        200: {"model": None},
+        202: {"description": "The invocation is queued"},
+        400: {"description": "The session has no invocations ready to invoke"},
+        404: {"description": "Session not found"},
+    },
+)
+async def invoke_session(
+    session_id: str = Path(description="The id of the session to invoke"),
+    all: bool = Query(
+        default=False, description="Whether or not to invoke all remaining invocations"
+    ),
+) -> None:
+    """Invokes a session"""
+    session = ApiDependencies.invoker.services.graph_execution_manager.get(session_id)
+    if session is None:
+        return Response(status_code=404)
+
+    if session.is_complete():
+        return Response(status_code=400)
+
+    ApiDependencies.invoker.invoke(session, invoke_all=all)
+    return Response(status_code=202)
+
+
+@session_router.delete(
+    "/{session_id}/invoke",
+    operation_id="cancel_session_invoke",
+    responses={
+        202: {"description": "The invocation is canceled"}
+    },
+)
+async def cancel_session_invoke(
+    session_id: str = Path(description="The id of the session to cancel"),
+) -> None:
+    """Invokes a session"""
+    ApiDependencies.invoker.cancel(session_id)
+    return Response(status_code=202)

invokeai/app/api/sockets.py

@@ -0,0 +1,38 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+from fastapi import FastAPI
+from fastapi_events.handlers.local import local_handler
+from fastapi_events.typing import Event
+from fastapi_socketio import SocketManager
+
+from ..services.events import EventServiceBase
+
+
+class SocketIO:
+    __sio: SocketManager
+
+    def __init__(self, app: FastAPI):
+        self.__sio = SocketManager(app=app)
+        self.__sio.on("subscribe", handler=self._handle_sub)
+        self.__sio.on("unsubscribe", handler=self._handle_unsub)
+
+        local_handler.register(
+            event_name=EventServiceBase.session_event, _func=self._handle_session_event
+        )
+
+    async def _handle_session_event(self, event: Event):
+        await self.__sio.emit(
+            event=event[1]["event"],
+            data=event[1]["data"],
+            room=event[1]["data"]["graph_execution_state_id"],
+        )
+
+    async def _handle_sub(self, sid, data, *args, **kwargs):
+        if "session" in data:
+            self.__sio.enter_room(sid, data["session"])
+
+        # @app.sio.on('unsubscribe')
+
+    async def _handle_unsub(self, sid, data, *args, **kwargs):
+        if "session" in data:
+            self.__sio.leave_room(sid, data["session"])

invokeai/app/api_app.py

@@ -0,0 +1,158 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+import asyncio
+from inspect import signature
+
+import uvicorn
+from fastapi import FastAPI
+from fastapi.middleware.cors import CORSMiddleware
+from fastapi.openapi.docs import get_redoc_html, get_swagger_ui_html
+from fastapi.openapi.utils import get_openapi
+from fastapi.staticfiles import StaticFiles
+from fastapi_events.handlers.local import local_handler
+from fastapi_events.middleware import EventHandlerASGIMiddleware
+from pydantic.schema import schema
+
+from ..backend import Args
+from .api.dependencies import ApiDependencies
+from .api.routers import images, sessions
+from .api.sockets import SocketIO
+from .invocations import *
+from .invocations.baseinvocation import BaseInvocation
+
+# Create the app
+# TODO: create this all in a method so configuration/etc. can be passed in?
+app = FastAPI(title="Invoke AI", docs_url=None, redoc_url=None)
+
+# Add event handler
+event_handler_id: int = id(app)
+app.add_middleware(
+    EventHandlerASGIMiddleware,
+    handlers=[
+        local_handler
+    ],  # TODO: consider doing this in services to support different configurations
+    middleware_id=event_handler_id,
+)
+
+# Add CORS
+# TODO: use configuration for this
+origins = []
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=origins,
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+socket_io = SocketIO(app)
+
+config = {}
+
+
+# Add startup event to load dependencies
+@app.on_event("startup")
+async def startup_event():
+    config = Args()
+    config.parse_args()
+
+    ApiDependencies.initialize(
+        config=config, event_handler_id=event_handler_id
+    )
+
+
+# Shut down threads
+@app.on_event("shutdown")
+async def shutdown_event():
+    ApiDependencies.shutdown()
+
+
+# Include all routers
+# TODO: REMOVE
+# app.include_router(
+#     invocation.invocation_router,
+#     prefix = '/api')
+
+app.include_router(sessions.session_router, prefix="/api")
+
+app.include_router(images.images_router, prefix="/api")
+
+
+# Build a custom OpenAPI to include all outputs
+# TODO: can outputs be included on metadata of invocation schemas somehow?
+def custom_openapi():
+    if app.openapi_schema:
+        return app.openapi_schema
+    openapi_schema = get_openapi(
+        title=app.title,
+        description="An API for invoking AI image operations",
+        version="1.0.0",
+        routes=app.routes,
+    )
+
+    # Add all outputs
+    all_invocations = BaseInvocation.get_invocations()
+    output_types = set()
+    output_type_titles = dict()
+    for invoker in all_invocations:
+        output_type = signature(invoker.invoke).return_annotation
+        output_types.add(output_type)
+
+    output_schemas = schema(output_types, ref_prefix="#/components/schemas/")
+    for schema_key, output_schema in output_schemas["definitions"].items():
+        openapi_schema["components"]["schemas"][schema_key] = output_schema
+
+        # TODO: note that we assume the schema_key here is the TYPE.__name__
+        # This could break in some cases, figure out a better way to do it
+        output_type_titles[schema_key] = output_schema["title"]
+
+    # Add a reference to the output type to additionalProperties of the invoker schema
+    for invoker in all_invocations:
+        invoker_name = invoker.__name__
+        output_type = signature(invoker.invoke).return_annotation
+        output_type_title = output_type_titles[output_type.__name__]
+        invoker_schema = openapi_schema["components"]["schemas"][invoker_name]
+        outputs_ref = {"$ref": f"#/components/schemas/{output_type_title}"}
+
+        invoker_schema["output"] = outputs_ref
+
+    app.openapi_schema = openapi_schema
+    return app.openapi_schema
+
+
+app.openapi = custom_openapi
+
+# Override API doc favicons
+app.mount("/static", StaticFiles(directory="static/dream_web"), name="static")
+
+
+@app.get("/docs", include_in_schema=False)
+def overridden_swagger():
+    return get_swagger_ui_html(
+        openapi_url=app.openapi_url,
+        title=app.title,
+        swagger_favicon_url="/static/favicon.ico",
+    )
+
+
+@app.get("/redoc", include_in_schema=False)
+def overridden_redoc():
+    return get_redoc_html(
+        openapi_url=app.openapi_url,
+        title=app.title,
+        redoc_favicon_url="/static/favicon.ico",
+    )
+
+
+def invoke_api():
+    # Start our own event loop for eventing usage
+    # TODO: determine if there's a better way to do this
+    loop = asyncio.new_event_loop()
+    config = uvicorn.Config(app=app, host="0.0.0.0", port=9090, loop=loop)
+    # Use access_log to turn off logging
+
+    server = uvicorn.Server(config)
+    loop.run_until_complete(server.serve())
+
+
+if __name__ == "__main__":
+    invoke_api()

invokeai/app/cli/commands.py

@@ -0,0 +1,202 @@
+# Copyright (c) 2023 Kyle Schouviller (https://github.com/kyle0654)
+
+from abc import ABC, abstractmethod
+import argparse
+from typing import Any, Callable, Iterable, Literal, get_args, get_origin, get_type_hints
+from pydantic import BaseModel, Field
+
+from ..invocations.image import ImageField
+from ..services.graph import GraphExecutionState
+from ..services.invoker import Invoker
+
+
+def add_parsers(
+    subparsers,
+    commands: list[type],
+    command_field: str = "type",
+    exclude_fields: list[str] = ["id", "type"],
+    add_arguments: Callable[[argparse.ArgumentParser], None]|None = None
+    ):
+    """Adds parsers for each command to the subparsers"""
+
+    # Create subparsers for each command
+    for command in commands:
+        hints = get_type_hints(command)
+        cmd_name = get_args(hints[command_field])[0]
+        command_parser = subparsers.add_parser(cmd_name, help=command.__doc__)
+
+        if add_arguments is not None:
+            add_arguments(command_parser)
+
+        # Convert all fields to arguments
+        fields = command.__fields__ # type: ignore
+        for name, field in fields.items():
+            if name in exclude_fields:
+                continue
+
+            if get_origin(field.type_) == Literal:
+                allowed_values = get_args(field.type_)
+                allowed_types = set()
+                for val in allowed_values:
+                    allowed_types.add(type(val))
+                allowed_types_list = list(allowed_types)
+                field_type = allowed_types_list[0] if len(allowed_types) == 1 else Union[allowed_types_list]  # type: ignore
+
+                command_parser.add_argument(
+                    f"--{name}",
+                    dest=name,
+                    type=field_type,
+                    default=field.default,
+                    choices=allowed_values,
+                    help=field.field_info.description,
+                )
+            else:
+                command_parser.add_argument(
+                    f"--{name}",
+                    dest=name,
+                    type=field.type_,
+                    default=field.default,
+                    help=field.field_info.description,
+                )
+
+
+class CliContext:
+    invoker: Invoker
+    session: GraphExecutionState
+    parser: argparse.ArgumentParser
+    defaults: dict[str, Any]
+
+    def __init__(self, invoker: Invoker, session: GraphExecutionState, parser: argparse.ArgumentParser):
+        self.invoker = invoker
+        self.session = session
+        self.parser = parser
+        self.defaults = dict()
+
+    def get_session(self):
+        self.session = self.invoker.services.graph_execution_manager.get(self.session.id)
+        return self.session
+
+
+class ExitCli(Exception):
+    """Exception to exit the CLI"""
+    pass
+
+
+class BaseCommand(ABC, BaseModel):
+    """A CLI command"""
+
+    # All commands must include a type name like this:
+    # type: Literal['your_command_name'] = 'your_command_name'
+
+    @classmethod
+    def get_all_subclasses(cls):
+        subclasses = []
+        toprocess = [cls]
+        while len(toprocess) > 0:
+            next = toprocess.pop(0)
+            next_subclasses = next.__subclasses__()
+            subclasses.extend(next_subclasses)
+            toprocess.extend(next_subclasses)
+        return subclasses
+
+    @classmethod
+    def get_commands(cls):
+        return tuple(BaseCommand.get_all_subclasses())
+
+    @classmethod
+    def get_commands_map(cls):
+        # Get the type strings out of the literals and into a dictionary
+        return dict(map(lambda t: (get_args(get_type_hints(t)['type'])[0], t),BaseCommand.get_all_subclasses()))
+
+    @abstractmethod
+    def run(self, context: CliContext) -> None:
+        """Run the command. Raise ExitCli to exit."""
+        pass
+
+
+class ExitCommand(BaseCommand):
+    """Exits the CLI"""
+    type: Literal['exit'] = 'exit'
+
+    def run(self, context: CliContext) -> None:
+        raise ExitCli()
+
+
+class HelpCommand(BaseCommand):
+    """Shows help"""
+    type: Literal['help'] = 'help'
+
+    def run(self, context: CliContext) -> None:
+        context.parser.print_help()
+
+
+def get_graph_execution_history(
+    graph_execution_state: GraphExecutionState,
+) -> Iterable[str]:
+    """Gets the history of fully-executed invocations for a graph execution"""
+    return (
+        n
+        for n in reversed(graph_execution_state.executed_history)
+        if n in graph_execution_state.graph.nodes
+    )
+
+
+def get_invocation_command(invocation) -> str:
+    fields = invocation.__fields__.items()
+    type_hints = get_type_hints(type(invocation))
+    command = [invocation.type]
+    for name, field in fields:
+        if name in ["id", "type"]:
+            continue
+
+        # TODO: add links
+
+        # Skip image fields when serializing command
+        type_hint = type_hints.get(name) or None
+        if type_hint is ImageField or ImageField in get_args(type_hint):
+            continue
+
+        field_value = getattr(invocation, name)
+        field_default = field.default
+        if field_value != field_default:
+            if type_hint is str or str in get_args(type_hint):
+                command.append(f'--{name} "{field_value}"')
+            else:
+                command.append(f"--{name} {field_value}")
+
+    return " ".join(command)
+
+
+class HistoryCommand(BaseCommand):
+    """Shows the invocation history"""
+    type: Literal['history'] = 'history'
+
+    # Inputs
+    # fmt: off
+    count: int = Field(default=5, gt=0, description="The number of history entries to show")
+    # fmt: on
+
+    def run(self, context: CliContext) -> None:
+        history = list(get_graph_execution_history(context.get_session()))
+        for i in range(min(self.count, len(history))):
+            entry_id = history[-1 - i]
+            entry = context.get_session().graph.get_node(entry_id)
+            print(f"{entry_id}: {get_invocation_command(entry)}")
+
+
+class SetDefaultCommand(BaseCommand):
+    """Sets a default value for a field"""
+    type: Literal['default'] = 'default'
+
+    # Inputs
+    # fmt: off
+    field: str = Field(description="The field to set the default for")
+    value: str = Field(description="The value to set the default to, or None to clear the default")
+    # fmt: on
+
+    def run(self, context: CliContext) -> None:
+        if self.value is None:
+            if self.field in context.defaults:
+                del context.defaults[self.field]
+        else:
+            context.defaults[self.field] = self.value

invokeai/app/cli_app.py

@@ -0,0 +1,275 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+import argparse
+import os
+import shlex
+import time
+from typing import (
+    Union,
+    get_type_hints,
+)
+
+from pydantic import BaseModel
+from pydantic.fields import Field
+
+from ..backend import Args
+from .cli.commands import BaseCommand, CliContext, ExitCli, add_parsers, get_graph_execution_history
+from .invocations import *
+from .invocations.baseinvocation import BaseInvocation
+from .services.events import EventServiceBase
+from .services.model_manager_initializer import get_model_manager
+from .services.restoration_services import RestorationServices
+from .services.graph import Edge, EdgeConnection, GraphExecutionState
+from .services.image_storage import DiskImageStorage
+from .services.invocation_queue import MemoryInvocationQueue
+from .services.invocation_services import InvocationServices
+from .services.invoker import Invoker
+from .services.processor import DefaultInvocationProcessor
+from .services.sqlite import SqliteItemStorage
+
+
+class CliCommand(BaseModel):
+    command: Union[BaseCommand.get_commands() + BaseInvocation.get_invocations()] = Field(discriminator="type")  # type: ignore
+
+
+class InvalidArgs(Exception):
+    pass
+
+
+def add_invocation_args(command_parser):
+    # Add linking capability
+    command_parser.add_argument(
+        "--link",
+        "-l",
+        action="append",
+        nargs=3,
+        help="A link in the format 'dest_field source_node source_field'. source_node can be relative to history (e.g. -1)",
+    )
+
+    command_parser.add_argument(
+        "--link_node",
+        "-ln",
+        action="append",
+        help="A link from all fields in the specified node. Node can be relative to history (e.g. -1)",
+    )
+
+
+def get_command_parser() -> argparse.ArgumentParser:
+    # Create invocation parser
+    parser = argparse.ArgumentParser()
+
+    def exit(*args, **kwargs):
+        raise InvalidArgs
+
+    parser.exit = exit
+    subparsers = parser.add_subparsers(dest="type")
+
+    # Create subparsers for each invocation
+    invocations = BaseInvocation.get_all_subclasses()
+    add_parsers(subparsers, invocations, add_arguments=add_invocation_args)
+
+    # Create subparsers for each command
+    commands = BaseCommand.get_all_subclasses()
+    add_parsers(subparsers, commands, exclude_fields=["type"])
+
+    return parser
+
+
+def generate_matching_edges(
+    a: BaseInvocation, b: BaseInvocation
+) -> list[Edge]:
+    """Generates all possible edges between two invocations"""
+    atype = type(a)
+    btype = type(b)
+
+    aoutputtype = atype.get_output_type()
+
+    afields = get_type_hints(aoutputtype)
+    bfields = get_type_hints(btype)
+
+    matching_fields = set(afields.keys()).intersection(bfields.keys())
+
+    # Remove invalid fields
+    invalid_fields = set(["type", "id"])
+    matching_fields = matching_fields.difference(invalid_fields)
+
+    edges = [
+        Edge(
+            source=EdgeConnection(node_id=a.id, field=field),
+            destination=EdgeConnection(node_id=b.id, field=field)
+        )
+        for field in matching_fields
+    ]
+    return edges
+
+
+class SessionError(Exception):
+    """Raised when a session error has occurred"""
+    pass
+
+
+def invoke_all(context: CliContext):
+    """Runs all invocations in the specified session"""
+    context.invoker.invoke(context.session, invoke_all=True)
+    while not context.get_session().is_complete():
+        # Wait some time
+        time.sleep(0.1)
+
+    # Print any errors
+    if context.session.has_error():
+        for n in context.session.errors:
+            print(
+                f"Error in node {n} (source node {context.session.prepared_source_mapping[n]}): {context.session.errors[n]}"
+            )
+        
+        raise SessionError()
+
+
+def invoke_cli():
+    config = Args()
+    config.parse_args()
+    model_manager = get_model_manager(config)
+
+    events = EventServiceBase()
+
+    output_folder = os.path.abspath(
+        os.path.join(os.path.dirname(__file__), "../../../outputs")
+    )
+
+    # TODO: build a file/path manager?
+    db_location = os.path.join(output_folder, "invokeai.db")
+
+    services = InvocationServices(
+        model_manager=model_manager,
+        events=events,
+        images=DiskImageStorage(output_folder),
+        queue=MemoryInvocationQueue(),
+        graph_execution_manager=SqliteItemStorage[GraphExecutionState](
+            filename=db_location, table_name="graph_executions"
+        ),
+        processor=DefaultInvocationProcessor(),
+        restoration=RestorationServices(config),
+    )
+
+    invoker = Invoker(services)
+    session: GraphExecutionState = invoker.create_execution_state()
+    parser = get_command_parser()
+
+    # Uncomment to print out previous sessions at startup
+    # print(services.session_manager.list())
+
+    context = CliContext(invoker, session, parser)
+
+    while True:
+        try:
+            cmd_input = input("> ")
+        except KeyboardInterrupt:
+            # Ctrl-c exits
+            break
+
+        try:
+            # Refresh the state of the session
+            history = list(get_graph_execution_history(context.session))
+
+            # Split the command for piping
+            cmds = cmd_input.split("|")
+            start_id = len(history)
+            current_id = start_id
+            new_invocations = list()
+            for cmd in cmds:
+                if cmd is None or cmd.strip() == "":
+                    raise InvalidArgs("Empty command")
+
+                # Parse args to create invocation
+                args = vars(context.parser.parse_args(shlex.split(cmd.strip())))
+
+                # Override defaults
+                for field_name, field_default in context.defaults.items():
+                    if field_name in args:
+                        args[field_name] = field_default
+
+                # Parse invocation
+                args["id"] = current_id
+                command = CliCommand(command=args)
+
+                # Run any CLI commands immediately
+                if isinstance(command.command, BaseCommand):
+                    # Invoke all current nodes to preserve operation order
+                    invoke_all(context)
+
+                    # Run the command
+                    command.command.run(context)
+                    continue
+
+                # Pipe previous command output (if there was a previous command)
+                edges: list[Edge] = list()
+                if len(history) > 0 or current_id != start_id:
+                    from_id = (
+                        history[0] if current_id == start_id else str(current_id - 1)
+                    )
+                    from_node = (
+                        next(filter(lambda n: n[0].id == from_id, new_invocations))[0]
+                        if current_id != start_id
+                        else context.session.graph.get_node(from_id)
+                    )
+                    matching_edges = generate_matching_edges(
+                        from_node, command.command
+                    )
+                    edges.extend(matching_edges)
+
+                # Parse provided links
+                if "link_node" in args and args["link_node"]:
+                    for link in args["link_node"]:
+                        link_node = context.session.graph.get_node(link)
+                        matching_edges = generate_matching_edges(
+                            link_node, command.command
+                        )
+                        matching_destinations = [e.destination for e in matching_edges]
+                        edges = [e for e in edges if e.destination not in matching_destinations]
+                        edges.extend(matching_edges)
+
+                if "link" in args and args["link"]:
+                    for link in args["link"]:
+                        edges = [e for e in edges if e.destination.node_id != command.command.id and e.destination.field != link[2]]
+                        edges.append(
+                            Edge(
+                                source=EdgeConnection(node_id=link[1], field=link[0]),
+                                destination=EdgeConnection(
+                                    node_id=command.command.id, field=link[2]
+                                )
+                            )
+                        )
+
+                new_invocations.append((command.command, edges))
+
+                current_id = current_id + 1
+
+                # Add the node to the session
+                context.session.add_node(command.command)
+                for edge in edges:
+                    print(edge)
+                    context.session.add_edge(edge)
+
+            # Execute all remaining nodes
+            invoke_all(context)
+
+        except InvalidArgs:
+            print('Invalid command, use "help" to list commands')
+            continue
+
+        except SessionError:
+            # Start a new session
+            print("Session error: creating a new session")
+            context.session = context.invoker.create_execution_state()
+
+        except ExitCli:
+            break
+
+        except SystemExit:
+            continue
+
+    invoker.stop()
+
+
+if __name__ == "__main__":
+    invoke_cli()

invokeai/app/invocations/__init__.py

@@ -0,0 +1,12 @@
+import os
+
+__all__ = []
+
+dirname = os.path.dirname(os.path.abspath(__file__))
+for f in os.listdir(dirname):
+    if (
+        f != "__init__.py"
+        and os.path.isfile("%s/%s" % (dirname, f))
+        and f[-3:] == ".py"
+    ):
+        __all__.append(f[:-3])

invokeai/app/invocations/baseinvocation.py

@@ -0,0 +1,78 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+from abc import ABC, abstractmethod
+from inspect import signature
+from typing import get_args, get_type_hints
+
+from pydantic import BaseModel, Field
+
+from ..services.invocation_services import InvocationServices
+
+
+class InvocationContext:
+    services: InvocationServices
+    graph_execution_state_id: str
+
+    def __init__(self, services: InvocationServices, graph_execution_state_id: str):
+        self.services = services
+        self.graph_execution_state_id = graph_execution_state_id
+
+
+class BaseInvocationOutput(BaseModel):
+    """Base class for all invocation outputs"""
+
+    # All outputs must include a type name like this:
+    # type: Literal['your_output_name']
+
+    @classmethod
+    def get_all_subclasses_tuple(cls):
+        subclasses = []
+        toprocess = [cls]
+        while len(toprocess) > 0:
+            next = toprocess.pop(0)
+            next_subclasses = next.__subclasses__()
+            subclasses.extend(next_subclasses)
+            toprocess.extend(next_subclasses)
+        return tuple(subclasses)
+
+
+class BaseInvocation(ABC, BaseModel):
+    """A node to process inputs and produce outputs.
+    May use dependency injection in __init__ to receive providers.
+    """
+
+    # All invocations must include a type name like this:
+    # type: Literal['your_output_name']
+
+    @classmethod
+    def get_all_subclasses(cls):
+        subclasses = []
+        toprocess = [cls]
+        while len(toprocess) > 0:
+            next = toprocess.pop(0)
+            next_subclasses = next.__subclasses__()
+            subclasses.extend(next_subclasses)
+            toprocess.extend(next_subclasses)
+        return subclasses
+
+    @classmethod
+    def get_invocations(cls):
+        return tuple(BaseInvocation.get_all_subclasses())
+
+    @classmethod
+    def get_invocations_map(cls):
+        # Get the type strings out of the literals and into a dictionary
+        return dict(map(lambda t: (get_args(get_type_hints(t)['type'])[0], t),BaseInvocation.get_all_subclasses()))
+    
+    @classmethod
+    def get_output_type(cls):
+        return signature(cls.invoke).return_annotation
+
+    @abstractmethod
+    def invoke(self, context: InvocationContext) -> BaseInvocationOutput:
+        """Invoke with provided context and return outputs."""
+        pass
+    
+    #fmt: off
+    id: str = Field(description="The id of this node. Must be unique among all nodes.")
+    #fmt: on

invokeai/app/invocations/cv.py

@@ -0,0 +1,50 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+from typing import Literal
+
+import cv2 as cv
+import numpy
+from PIL import Image, ImageOps
+from pydantic import Field
+
+from ..services.image_storage import ImageType
+from .baseinvocation import BaseInvocation, InvocationContext
+from .image import ImageField, ImageOutput
+
+
+class CvInpaintInvocation(BaseInvocation):
+    """Simple inpaint using opencv."""
+    #fmt: off
+    type: Literal["cv_inpaint"] = "cv_inpaint"
+
+    # Inputs
+    image: ImageField = Field(default=None, description="The image to inpaint")
+    mask: ImageField = Field(default=None, description="The mask to use when inpainting")
+    #fmt: on
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = context.services.images.get(
+            self.image.image_type, self.image.image_name
+        )
+        mask = context.services.images.get(self.mask.image_type, self.mask.image_name)
+
+        # Convert to cv image/mask
+        # TODO: consider making these utility functions
+        cv_image = cv.cvtColor(numpy.array(image.convert("RGB")), cv.COLOR_RGB2BGR)
+        cv_mask = numpy.array(ImageOps.invert(mask))
+
+        # Inpaint
+        cv_inpainted = cv.inpaint(cv_image, cv_mask, 3, cv.INPAINT_TELEA)
+
+        # Convert back to Pillow
+        # TODO: consider making a utility function
+        image_inpainted = Image.fromarray(cv.cvtColor(cv_inpainted, cv.COLOR_BGR2RGB))
+
+        image_type = ImageType.INTERMEDIATE
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, image_inpainted)
+        return ImageOutput(
+            image=ImageField(image_type=image_type, image_name=image_name)
+        )

invokeai/app/invocations/generate.py

@@ -0,0 +1,221 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+from datetime import datetime, timezone
+from typing import Any, Literal, Optional, Union
+
+import numpy as np
+
+from torch import Tensor
+from PIL import Image
+from pydantic import Field
+from skimage.exposure.histogram_matching import match_histograms
+
+from ..services.image_storage import ImageType
+from ..services.invocation_services import InvocationServices
+from .baseinvocation import BaseInvocation, InvocationContext
+from .image import ImageField, ImageOutput
+from ...backend.generator import Txt2Img, Img2Img, Inpaint, InvokeAIGenerator, Generator
+from ...backend.stable_diffusion import PipelineIntermediateState
+from ...backend.util.util import image_to_dataURL
+
+SAMPLER_NAME_VALUES = Literal[
+    tuple(InvokeAIGenerator.schedulers())
+]
+
+# Text to image
+class TextToImageInvocation(BaseInvocation):
+    """Generates an image using text2img."""
+
+    type: Literal["txt2img"] = "txt2img"
+
+    # Inputs
+    # TODO: consider making prompt optional to enable providing prompt through a link
+    # fmt: off
+    prompt: Optional[str] = Field(description="The prompt to generate an image from")
+    seed:        int = Field(default=-1,ge=-1, le=np.iinfo(np.uint32).max, description="The seed to use (-1 for a random seed)", )
+    steps:       int = Field(default=10, gt=0, description="The number of steps to use to generate the image")
+    width:       int = Field(default=512, multiple_of=64, gt=0, description="The width of the resulting image", )
+    height:      int = Field(default=512, multiple_of=64, gt=0, description="The height of the resulting image", )
+    cfg_scale: float = Field(default=7.5, gt=0, description="The Classifier-Free Guidance, higher values may result in a result closer to the prompt", )
+    sampler_name: SAMPLER_NAME_VALUES = Field(default="k_lms", description="The sampler to use" )
+    seamless:   bool = Field(default=False, description="Whether or not to generate an image that can tile without seams", )
+    model:       str = Field(default="", description="The model to use (currently ignored)")
+    progress_images: bool = Field(default=False, description="Whether or not to produce progress images during generation",  )
+    # fmt: on
+
+    # TODO: pass this an emitter method or something? or a session for dispatching?
+    def dispatch_progress(
+        self, context: InvocationContext, sample: Tensor, step: int
+    ) -> None:  
+        # TODO: only output a preview image when requested
+        image = Generator.sample_to_lowres_estimated_image(sample)
+
+        (width, height) = image.size
+        width *= 8
+        height *= 8
+
+        dataURL = image_to_dataURL(image, image_format="JPEG")
+
+        context.services.events.emit_generator_progress(
+            context.graph_execution_state_id,
+            self.id,
+            {
+                "width": width,
+                "height": height,
+                "dataURL": dataURL
+            },
+            step,
+            self.steps,
+        )
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        def step_callback(state: PipelineIntermediateState):
+            self.dispatch_progress(context, state.latents, state.step)
+
+        # Handle invalid model parameter
+        # TODO: figure out if this can be done via a validator that uses the model_cache
+        # TODO: How to get the default model name now?
+        #       (right now uses whatever current model is set in model manager)
+        model= context.services.model_manager.get_model()
+        outputs = Txt2Img(model).generate(
+            prompt=self.prompt,
+            step_callback=step_callback,
+            **self.dict(
+                exclude={"prompt"}
+            ),  # Shorthand for passing all of the parameters above manually
+        )
+        # Outputs is an infinite iterator that will return a new InvokeAIGeneratorOutput object
+        # each time it is called. We only need the first one.
+        generate_output = next(outputs)
+
+        # Results are image and seed, unwrap for now and ignore the seed
+        # TODO: pre-seed?
+        # TODO: can this return multiple results? Should it?
+        image_type = ImageType.RESULT
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, generate_output.image)
+        return ImageOutput(
+            image=ImageField(image_type=image_type, image_name=image_name)
+        )
+
+
+class ImageToImageInvocation(TextToImageInvocation):
+    """Generates an image using img2img."""
+
+    type: Literal["img2img"] = "img2img"
+
+    # Inputs
+    image: Union[ImageField, None] = Field(description="The input image")
+    strength: float = Field(
+        default=0.75, gt=0, le=1, description="The strength of the original image"
+    )
+    fit: bool = Field(
+        default=True,
+        description="Whether or not the result should be fit to the aspect ratio of the input image",
+    )
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = (
+            None
+            if self.image is None
+            else context.services.images.get(
+                self.image.image_type, self.image.image_name
+            )
+        )
+        mask = None
+
+        def step_callback(sample, step=0):
+            self.dispatch_progress(context, sample, step)
+
+        # Handle invalid model parameter
+        # TODO: figure out if this can be done via a validator that uses the model_cache
+        # TODO: How to get the default model name now?
+        model = context.services.model_manager.get_model()
+        generator_output = next(
+            Img2Img(model).generate(
+                prompt=self.prompt,
+                init_image=image,
+                init_mask=mask,
+                step_callback=step_callback,
+                **self.dict(
+                    exclude={"prompt", "image", "mask"}
+                ),  # Shorthand for passing all of the parameters above manually
+            )
+        )
+
+        result_image = generator_output.image
+
+        # Results are image and seed, unwrap for now and ignore the seed
+        # TODO: pre-seed?
+        # TODO: can this return multiple results? Should it?
+        image_type = ImageType.RESULT
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, result_image)
+        return ImageOutput(
+            image=ImageField(image_type=image_type, image_name=image_name)
+        )
+
+class InpaintInvocation(ImageToImageInvocation):
+    """Generates an image using inpaint."""
+
+    type: Literal["inpaint"] = "inpaint"
+
+    # Inputs
+    mask: Union[ImageField, None] = Field(description="The mask")
+    inpaint_replace: float = Field(
+        default=0.0,
+        ge=0.0,
+        le=1.0,
+        description="The amount by which to replace masked areas with latent noise",
+    )
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = (
+            None
+            if self.image is None
+            else context.services.images.get(
+                self.image.image_type, self.image.image_name
+            )
+        )
+        mask = (
+            None
+            if self.mask is None
+            else context.services.images.get(self.mask.image_type, self.mask.image_name)
+        )
+
+        def step_callback(sample, step=0):
+            self.dispatch_progress(context, sample, step)
+
+        # Handle invalid model parameter
+        # TODO: figure out if this can be done via a validator that uses the model_cache
+        # TODO: How to get the default model name now?
+        manager = context.services.model_manager.get_model()
+        generator_output = next(
+            Inpaint(model).generate(
+                prompt=self.prompt,
+                init_image=image,
+                mask_image=mask,
+                step_callback=step_callback,
+                **self.dict(
+                    exclude={"prompt", "image", "mask"}
+                ),  # Shorthand for passing all of the parameters above manually
+            )
+        )
+
+        result_image = generator_output.image
+
+        # Results are image and seed, unwrap for now and ignore the seed
+        # TODO: pre-seed?
+        # TODO: can this return multiple results? Should it?
+        image_type = ImageType.RESULT
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, result_image)
+        return ImageOutput(
+            image=ImageField(image_type=image_type, image_name=image_name)
+        )

invokeai/app/invocations/image.py

@@ -0,0 +1,287 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+from datetime import datetime, timezone
+from typing import Literal, Optional
+
+import numpy
+from PIL import Image, ImageFilter, ImageOps
+from pydantic import BaseModel, Field
+
+from ..services.image_storage import ImageType
+from ..services.invocation_services import InvocationServices
+from .baseinvocation import BaseInvocation, BaseInvocationOutput, InvocationContext
+
+
+class ImageField(BaseModel):
+    """An image field used for passing image objects between invocations"""
+
+    image_type: str = Field(
+        default=ImageType.RESULT, description="The type of the image"
+    )
+    image_name: Optional[str] = Field(default=None, description="The name of the image")
+
+
+class ImageOutput(BaseInvocationOutput):
+    """Base class for invocations that output an image"""
+    #fmt: off
+    type: Literal["image"] = "image"
+    image:      ImageField = Field(default=None, description="The output image")
+    #fmt: on
+
+class MaskOutput(BaseInvocationOutput):
+    """Base class for invocations that output a mask"""
+    #fmt: off
+    type: Literal["mask"] = "mask"
+    mask:      ImageField = Field(default=None, description="The output mask")
+    #fomt: on
+
+# TODO: this isn't really necessary anymore
+class LoadImageInvocation(BaseInvocation):
+    """Load an image from a filename and provide it as output."""
+    #fmt: off
+    type: Literal["load_image"] = "load_image"
+
+    # Inputs
+    image_type: ImageType = Field(description="The type of the image")
+    image_name:       str = Field(description="The name of the image")
+    #fmt: on
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        return ImageOutput(
+            image=ImageField(image_type=self.image_type, image_name=self.image_name)
+        )
+
+
+class ShowImageInvocation(BaseInvocation):
+    """Displays a provided image, and passes it forward in the pipeline."""
+
+    type: Literal["show_image"] = "show_image"
+
+    # Inputs
+    image: ImageField = Field(default=None, description="The image to show")
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = context.services.images.get(
+            self.image.image_type, self.image.image_name
+        )
+        if image:
+            image.show()
+
+        # TODO: how to handle failure?
+
+        return ImageOutput(
+            image=ImageField(
+                image_type=self.image.image_type, image_name=self.image.image_name
+            )
+        )
+
+
+class CropImageInvocation(BaseInvocation):
+    """Crops an image to a specified box. The box can be outside of the image."""
+    #fmt: off
+    type: Literal["crop"] = "crop"
+
+    # Inputs
+    image: ImageField = Field(default=None, description="The image to crop")
+    x:      int = Field(default=0, description="The left x coordinate of the crop rectangle")
+    y:      int = Field(default=0, description="The top y coordinate of the crop rectangle")
+    width:  int = Field(default=512, gt=0, description="The width of the crop rectangle")
+    height: int = Field(default=512, gt=0, description="The height of the crop rectangle")
+    #fmt: on
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = context.services.images.get(
+            self.image.image_type, self.image.image_name
+        )
+
+        image_crop = Image.new(
+            mode="RGBA", size=(self.width, self.height), color=(0, 0, 0, 0)
+        )
+        image_crop.paste(image, (-self.x, -self.y))
+
+        image_type = ImageType.INTERMEDIATE
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, image_crop)
+        return ImageOutput(
+            image=ImageField(image_type=image_type, image_name=image_name)
+        )
+
+
+class PasteImageInvocation(BaseInvocation):
+    """Pastes an image into another image."""
+    #fmt: off
+    type: Literal["paste"] = "paste"
+
+    # Inputs
+    base_image:     ImageField = Field(default=None, description="The base image")
+    image:          ImageField = Field(default=None, description="The image to paste")
+    mask: Optional[ImageField] = Field(default=None, description="The mask to use when pasting")
+    x:                     int = Field(default=0, description="The left x coordinate at which to paste the image")
+    y:                     int = Field(default=0, description="The top y coordinate at which to paste the image")
+    #fmt: on
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        base_image = context.services.images.get(
+            self.base_image.image_type, self.base_image.image_name
+        )
+        image = context.services.images.get(
+            self.image.image_type, self.image.image_name
+        )
+        mask = (
+            None
+            if self.mask is None
+            else ImageOps.invert(
+                services.images.get(self.mask.image_type, self.mask.image_name)
+            )
+        )
+        # TODO: probably shouldn't invert mask here... should user be required to do it?
+
+        min_x = min(0, self.x)
+        min_y = min(0, self.y)
+        max_x = max(base_image.width, image.width + self.x)
+        max_y = max(base_image.height, image.height + self.y)
+
+        new_image = Image.new(
+            mode="RGBA", size=(max_x - min_x, max_y - min_y), color=(0, 0, 0, 0)
+        )
+        new_image.paste(base_image, (abs(min_x), abs(min_y)))
+        new_image.paste(image, (max(0, self.x), max(0, self.y)), mask=mask)
+
+        image_type = ImageType.RESULT
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, new_image)
+        return ImageOutput(
+            image=ImageField(image_type=image_type, image_name=image_name)
+        )
+
+
+class MaskFromAlphaInvocation(BaseInvocation):
+    """Extracts the alpha channel of an image as a mask."""
+    #fmt: off
+    type: Literal["tomask"] = "tomask"
+
+    # Inputs
+    image: ImageField = Field(default=None, description="The image to create the mask from")
+    invert:      bool = Field(default=False, description="Whether or not to invert the mask")
+    #fmt: on
+
+    def invoke(self, context: InvocationContext) -> MaskOutput:
+        image = context.services.images.get(
+            self.image.image_type, self.image.image_name
+        )
+
+        image_mask = image.split()[-1]
+        if self.invert:
+            image_mask = ImageOps.invert(image_mask)
+
+        image_type = ImageType.INTERMEDIATE
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, image_mask)
+        return MaskOutput(mask=ImageField(image_type=image_type, image_name=image_name))
+
+
+class BlurInvocation(BaseInvocation):
+    """Blurs an image"""
+
+    #fmt: off
+    type: Literal["blur"] = "blur"
+
+    # Inputs
+    image: ImageField = Field(default=None, description="The image to blur")
+    radius:     float = Field(default=8.0, ge=0, description="The blur radius")
+    blur_type: Literal["gaussian", "box"] = Field(default="gaussian", description="The type of blur")
+    #fmt: on
+    
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = context.services.images.get(
+            self.image.image_type, self.image.image_name
+        )
+
+        blur = (
+            ImageFilter.GaussianBlur(self.radius)
+            if self.blur_type == "gaussian"
+            else ImageFilter.BoxBlur(self.radius)
+        )
+        blur_image = image.filter(blur)
+
+        image_type = ImageType.INTERMEDIATE
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, blur_image)
+        return ImageOutput(
+            image=ImageField(image_type=image_type, image_name=image_name)
+        )
+
+
+class LerpInvocation(BaseInvocation):
+    """Linear interpolation of all pixels of an image"""
+    #fmt: off
+    type: Literal["lerp"] = "lerp"
+
+    # Inputs
+    image: ImageField = Field(default=None, description="The image to lerp")
+    min: int = Field(default=0, ge=0, le=255, description="The minimum output value")
+    max: int = Field(default=255, ge=0, le=255, description="The maximum output value")
+    #fmt: on
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = context.services.images.get(
+            self.image.image_type, self.image.image_name
+        )
+
+        image_arr = numpy.asarray(image, dtype=numpy.float32) / 255
+        image_arr = image_arr * (self.max - self.min) + self.max
+
+        lerp_image = Image.fromarray(numpy.uint8(image_arr))
+
+        image_type = ImageType.INTERMEDIATE
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, lerp_image)
+        return ImageOutput(
+            image=ImageField(image_type=image_type, image_name=image_name)
+        )
+
+
+class InverseLerpInvocation(BaseInvocation):
+    """Inverse linear interpolation of all pixels of an image"""
+    #fmt: off
+    type: Literal["ilerp"] = "ilerp"
+
+    # Inputs
+    image: ImageField = Field(default=None, description="The image to lerp")
+    min: int = Field(default=0, ge=0, le=255, description="The minimum input value")
+    max: int = Field(default=255, ge=0, le=255, description="The maximum input value")
+    #fmt: on
+    
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = context.services.images.get(
+            self.image.image_type, self.image.image_name
+        )
+
+        image_arr = numpy.asarray(image, dtype=numpy.float32)
+        image_arr = (
+            numpy.minimum(
+                numpy.maximum(image_arr - self.min, 0) / float(self.max - self.min), 1
+            )
+            * 255
+        )
+
+        ilerp_image = Image.fromarray(numpy.uint8(image_arr))
+
+        image_type = ImageType.INTERMEDIATE
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, ilerp_image)
+        return ImageOutput(
+            image=ImageField(image_type=image_type, image_name=image_name)
+        )

invokeai/app/invocations/prompt.py

@@ -0,0 +1,14 @@
+from typing import Literal
+
+from pydantic.fields import Field
+
+from .baseinvocation import BaseInvocationOutput
+
+
+class PromptOutput(BaseInvocationOutput):
+    """Base class for invocations that output a prompt"""
+    #fmt: off
+    type: Literal["prompt"] = "prompt"
+
+    prompt: str = Field(default=None, description="The output prompt")
+    #fmt: on

invokeai/app/invocations/reconstruct.py

@@ -0,0 +1,42 @@
+from datetime import datetime, timezone
+from typing import Literal, Union
+
+from pydantic import Field
+
+from ..services.image_storage import ImageType
+from ..services.invocation_services import InvocationServices
+from .baseinvocation import BaseInvocation, InvocationContext
+from .image import ImageField, ImageOutput
+
+class RestoreFaceInvocation(BaseInvocation):
+    """Restores faces in an image."""
+    #fmt: off
+    type:  Literal["restore_face"] = "restore_face"
+
+    # Inputs
+    image: Union[ImageField, None] = Field(description="The input image")
+    strength:                float = Field(default=0.75, gt=0, le=1, description="The strength of the restoration"  )
+    #fmt: on
+    
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = context.services.images.get(
+            self.image.image_type, self.image.image_name
+        )
+        results = context.services.restoration.upscale_and_reconstruct(
+            image_list=[[image, 0]],
+            upscale=None,
+            strength=self.strength,  # GFPGAN strength
+            save_original=False,
+            image_callback=None,
+        )
+
+        # Results are image and seed, unwrap for now
+        # TODO: can this return multiple results?
+        image_type = ImageType.RESULT
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, results[0][0])
+        return ImageOutput(
+            image=ImageField(image_type=image_type, image_name=image_name)
+        )

invokeai/app/invocations/upscale.py

@@ -0,0 +1,46 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+from datetime import datetime, timezone
+from typing import Literal, Union
+
+from pydantic import Field
+
+from ..services.image_storage import ImageType
+from ..services.invocation_services import InvocationServices
+from .baseinvocation import BaseInvocation, InvocationContext
+from .image import ImageField, ImageOutput
+
+
+class UpscaleInvocation(BaseInvocation):
+    """Upscales an image."""
+    #fmt: off
+    type: Literal["upscale"] = "upscale"
+
+    # Inputs
+    image: Union[ImageField, None] = Field(description="The input image", default=None)
+    strength: float = Field(default=0.75, gt=0, le=1, description="The strength")
+    level: Literal[2, 4] = Field(default=2, description="The upscale level")
+    #fmt: on
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = context.services.images.get(
+            self.image.image_type, self.image.image_name
+        )
+        results = context.services.restoration.upscale_and_reconstruct(
+            image_list=[[image, 0]],
+            upscale=(self.level, self.strength),
+            strength=0.0,  # GFPGAN strength
+            save_original=False,
+            image_callback=None,
+        )
+
+        # Results are image and seed, unwrap for now
+        # TODO: can this return multiple results?
+        image_type = ImageType.RESULT
+        image_name = context.services.images.create_name(
+            context.graph_execution_state_id, self.id
+        )
+        context.services.images.save(image_type, image_name, results[0][0])
+        return ImageOutput(
+            image=ImageField(image_type=image_type, image_name=image_name)
+        )

invokeai/app/services/events.py

@@ -0,0 +1,88 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+from typing import Any, Dict, TypedDict
+
+ProgressImage = TypedDict(
+    "ProgressImage", {"dataURL": str, "width": int, "height": int}
+)
+
+class EventServiceBase:
+    session_event: str = "session_event"
+
+    """Basic event bus, to have an empty stand-in when not needed"""
+
+    def dispatch(self, event_name: str, payload: Any) -> None:
+        pass
+
+    def __emit_session_event(self, event_name: str, payload: Dict) -> None:
+        self.dispatch(
+            event_name=EventServiceBase.session_event,
+            payload=dict(event=event_name, data=payload),
+        )
+
+    # Define events here for every event in the system.
+    # This will make them easier to integrate until we find a schema generator.
+    def emit_generator_progress(
+        self,
+        graph_execution_state_id: str,
+        invocation_id: str,
+        progress_image: ProgressImage | None,
+        step: int,
+        total_steps: int,
+    ) -> None:
+        """Emitted when there is generation progress"""
+        self.__emit_session_event(
+            event_name="generator_progress",
+            payload=dict(
+                graph_execution_state_id=graph_execution_state_id,
+                invocation_id=invocation_id,
+                progress_image=progress_image,
+                step=step,
+                total_steps=total_steps,
+            ),
+        )
+
+    def emit_invocation_complete(
+        self, graph_execution_state_id: str, invocation_id: str, result: Dict
+    ) -> None:
+        """Emitted when an invocation has completed"""
+        self.__emit_session_event(
+            event_name="invocation_complete",
+            payload=dict(
+                graph_execution_state_id=graph_execution_state_id,
+                invocation_id=invocation_id,
+                result=result,
+            ),
+        )
+
+    def emit_invocation_error(
+        self, graph_execution_state_id: str, invocation_id: str, error: str
+    ) -> None:
+        """Emitted when an invocation has completed"""
+        self.__emit_session_event(
+            event_name="invocation_error",
+            payload=dict(
+                graph_execution_state_id=graph_execution_state_id,
+                invocation_id=invocation_id,
+                error=error,
+            ),
+        )
+
+    def emit_invocation_started(
+        self, graph_execution_state_id: str, invocation_id: str
+    ) -> None:
+        """Emitted when an invocation has started"""
+        self.__emit_session_event(
+            event_name="invocation_started",
+            payload=dict(
+                graph_execution_state_id=graph_execution_state_id,
+                invocation_id=invocation_id,
+            ),
+        )
+
+    def emit_graph_execution_complete(self, graph_execution_state_id: str) -> None:
+        """Emitted when a session has completed all invocations"""
+        self.__emit_session_event(
+            event_name="graph_execution_state_complete",
+            payload=dict(graph_execution_state_id=graph_execution_state_id),
+        )

invokeai/app/services/image_storage.py

@@ -0,0 +1,113 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+import datetime
+import os
+from abc import ABC, abstractmethod
+from enum import Enum
+from pathlib import Path
+from queue import Queue
+from typing import Dict
+
+from PIL.Image import Image
+
+from invokeai.backend.image_util import PngWriter
+
+
+class ImageType(str, Enum):
+    RESULT = "results"
+    INTERMEDIATE = "intermediates"
+    UPLOAD = "uploads"
+
+
+class ImageStorageBase(ABC):
+    """Responsible for storing and retrieving images."""
+
+    @abstractmethod
+    def get(self, image_type: ImageType, image_name: str) -> Image:
+        pass
+
+    # TODO: make this a bit more flexible for e.g. cloud storage
+    @abstractmethod
+    def get_path(self, image_type: ImageType, image_name: str) -> str:
+        pass
+
+    @abstractmethod
+    def save(self, image_type: ImageType, image_name: str, image: Image) -> None:
+        pass
+
+    @abstractmethod
+    def delete(self, image_type: ImageType, image_name: str) -> None:
+        pass
+
+    def create_name(self, context_id: str, node_id: str) -> str:
+        return f"{context_id}_{node_id}_{str(int(datetime.datetime.now(datetime.timezone.utc).timestamp()))}.png"
+
+
+class DiskImageStorage(ImageStorageBase):
+    """Stores images on disk"""
+
+    __output_folder: str
+    __pngWriter: PngWriter
+    __cache_ids: Queue  # TODO: this is an incredibly naive cache
+    __cache: Dict[str, Image]
+    __max_cache_size: int
+
+    def __init__(self, output_folder: str):
+        self.__output_folder = output_folder
+        self.__pngWriter = PngWriter(output_folder)
+        self.__cache = dict()
+        self.__cache_ids = Queue()
+        self.__max_cache_size = 10  # TODO: get this from config
+
+        Path(output_folder).mkdir(parents=True, exist_ok=True)
+
+        # TODO: don't hard-code. get/save/delete should maybe take subpath?
+        for image_type in ImageType:
+            Path(os.path.join(output_folder, image_type)).mkdir(
+                parents=True, exist_ok=True
+            )
+
+    def get(self, image_type: ImageType, image_name: str) -> Image:
+        image_path = self.get_path(image_type, image_name)
+        cache_item = self.__get_cache(image_path)
+        if cache_item:
+            return cache_item
+
+        image = Image.open(image_path)
+        self.__set_cache(image_path, image)
+        return image
+
+    # TODO: make this a bit more flexible for e.g. cloud storage
+    def get_path(self, image_type: ImageType, image_name: str) -> str:
+        path = os.path.join(self.__output_folder, image_type, image_name)
+        return path
+
+    def save(self, image_type: ImageType, image_name: str, image: Image) -> None:
+        image_subpath = os.path.join(image_type, image_name)
+        self.__pngWriter.save_image_and_prompt_to_png(
+            image, "", image_subpath, None
+        )  # TODO: just pass full path to png writer
+
+        image_path = self.get_path(image_type, image_name)
+        self.__set_cache(image_path, image)
+
+    def delete(self, image_type: ImageType, image_name: str) -> None:
+        image_path = self.get_path(image_type, image_name)
+        if os.path.exists(image_path):
+            os.remove(image_path)
+
+        if image_path in self.__cache:
+            del self.__cache[image_path]
+
+    def __get_cache(self, image_name: str) -> Image:
+        return None if image_name not in self.__cache else self.__cache[image_name]
+
+    def __set_cache(self, image_name: str, image: Image):
+        if not image_name in self.__cache:
+            self.__cache[image_name] = image
+            self.__cache_ids.put(
+                image_name
+            )  # TODO: this should refresh position for LRU cache
+            if len(self.__cache) > self.__max_cache_size:
+                cache_id = self.__cache_ids.get()
+                del self.__cache[cache_id]

invokeai/app/services/invocation_queue.py

@@ -0,0 +1,81 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+from abc import ABC, abstractmethod
+from queue import Queue
+import time
+
+
+# TODO: make this serializable
+class InvocationQueueItem:
+    # session_id: str
+    graph_execution_state_id: str
+    invocation_id: str
+    invoke_all: bool
+    timestamp: float
+
+    def __init__(
+        self,
+        # session_id: str,
+        graph_execution_state_id: str,
+        invocation_id: str,
+        invoke_all: bool = False,
+    ):
+        # self.session_id = session_id
+        self.graph_execution_state_id = graph_execution_state_id
+        self.invocation_id = invocation_id
+        self.invoke_all = invoke_all
+        self.timestamp = time.time()
+
+
+class InvocationQueueABC(ABC):
+    """Abstract base class for all invocation queues"""
+
+    @abstractmethod
+    def get(self) -> InvocationQueueItem:
+        pass
+
+    @abstractmethod
+    def put(self, item: InvocationQueueItem | None) -> None:
+        pass
+
+    @abstractmethod
+    def cancel(self, graph_execution_state_id: str) -> None:
+        pass
+
+    @abstractmethod
+    def is_canceled(self, graph_execution_state_id: str) -> bool:
+        pass
+
+
+class MemoryInvocationQueue(InvocationQueueABC):
+    __queue: Queue
+    __cancellations: dict[str, float]
+
+    def __init__(self):
+        self.__queue = Queue()
+        self.__cancellations = dict()
+
+    def get(self) -> InvocationQueueItem:
+        item = self.__queue.get()
+
+        while isinstance(item, InvocationQueueItem) \
+            and item.graph_execution_state_id in self.__cancellations \
+            and self.__cancellations[item.graph_execution_state_id] > item.timestamp:
+            item = self.__queue.get()
+
+        # Clear old items
+        for graph_execution_state_id in list(self.__cancellations.keys()):
+            if self.__cancellations[graph_execution_state_id] < item.timestamp:
+                del self.__cancellations[graph_execution_state_id]
+
+        return item
+
+    def put(self, item: InvocationQueueItem | None) -> None:
+        self.__queue.put(item)
+
+    def cancel(self, graph_execution_state_id: str) -> None:
+        if graph_execution_state_id not in self.__cancellations:
+            self.__cancellations[graph_execution_state_id] = time.time()
+
+    def is_canceled(self, graph_execution_state_id: str) -> bool:
+        return graph_execution_state_id in self.__cancellations

invokeai/app/services/invocation_services.py

@@ -0,0 +1,39 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+from invokeai.backend import ModelManager
+
+from .events import EventServiceBase
+from .image_storage import ImageStorageBase
+from .restoration_services import RestorationServices
+from .invocation_queue import InvocationQueueABC
+from .item_storage import ItemStorageABC
+
+class InvocationServices:
+    """Services that can be used by invocations"""
+
+    events: EventServiceBase
+    images: ImageStorageBase
+    queue: InvocationQueueABC
+    model_manager: ModelManager
+    restoration: RestorationServices
+
+    # NOTE: we must forward-declare any types that include invocations, since invocations can use services
+    graph_execution_manager: ItemStorageABC["GraphExecutionState"]
+    processor: "InvocationProcessorABC"
+
+    def __init__(
+            self,
+            model_manager: ModelManager,
+            events: EventServiceBase,
+            images: ImageStorageBase,
+            queue: InvocationQueueABC,
+            graph_execution_manager: ItemStorageABC["GraphExecutionState"],
+            processor: "InvocationProcessorABC",
+            restoration: RestorationServices,
+    ):
+        self.model_manager = model_manager
+        self.events = events
+        self.images = images
+        self.queue = queue
+        self.graph_execution_manager = graph_execution_manager
+        self.processor = processor
+        self.restoration = restoration

invokeai/app/services/invoker.py

@@ -0,0 +1,91 @@
+# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
+
+from abc import ABC
+from threading import Event, Thread
+
+from ..invocations.baseinvocation import InvocationContext
+from .graph import Graph, GraphExecutionState
+from .invocation_queue import InvocationQueueABC, InvocationQueueItem
+from .invocation_services import InvocationServices
+from .item_storage import ItemStorageABC
+
+
+class Invoker:
+    """The invoker, used to execute invocations"""
+
+    services: InvocationServices
+
+    def __init__(self, services: InvocationServices):
+        self.services = services
+        self._start()
+
+    def invoke(
+        self, graph_execution_state: GraphExecutionState, invoke_all: bool = False
+    ) -> str | None:
+        """Determines the next node to invoke and returns the id of the invoked node, or None if there are no nodes to execute"""
+
+        # Get the next invocation
+        invocation = graph_execution_state.next()
+        if not invocation:
+            return None
+
+        # Save the execution state
+        self.services.graph_execution_manager.set(graph_execution_state)
+
+        # Queue the invocation
+        print(f"queueing item {invocation.id}")
+        self.services.queue.put(
+            InvocationQueueItem(
+                # session_id    = session.id,
+                graph_execution_state_id=graph_execution_state.id,
+                invocation_id=invocation.id,
+                invoke_all=invoke_all,
+            )
+        )
+
+        return invocation.id
+
+    def create_execution_state(self, graph: Graph | None = None) -> GraphExecutionState:
+        """Creates a new execution state for the given graph"""
+        new_state = GraphExecutionState(graph=Graph() if graph is None else graph)
+        self.services.graph_execution_manager.set(new_state)
+        return new_state
+    
+    def cancel(self, graph_execution_state_id: str) -> None:
+        """Cancels the given execution state"""
+        self.services.queue.cancel(graph_execution_state_id)
+
+    def __start_service(self, service) -> None:
+        # Call start() method on any services that have it
+        start_op = getattr(service, "start", None)
+        if callable(start_op):
+            start_op(self)
+
+    def __stop_service(self, service) -> None:
+        # Call stop() method on any services that have it
+        stop_op = getattr(service, "stop", None)
+        if callable(stop_op):
+            stop_op(self)
+
+    def _start(self) -> None:
+        """Starts the invoker. This is called automatically when the invoker is created."""
+        for service in vars(self.services):
+            self.__start_service(getattr(self.services, service))
+
+        for service in vars(self.services):
+            self.__start_service(getattr(self.services, service))
+
+    def stop(self) -> None:
+        """Stops the invoker. A new invoker will have to be created to execute further."""
+        # First stop all services
+        for service in vars(self.services):
+            self.__stop_service(getattr(self.services, service))
+
+        for service in vars(self.services):
+            self.__stop_service(getattr(self.services, service))
+
+        self.services.queue.put(None)
+
+
+class InvocationProcessorABC(ABC):
+    pass

invokeai/app/services/item_storage.py

@@ -0,0 +1,62 @@
+from abc import ABC, abstractmethod
+from typing import Callable, Generic, TypeVar
+
+from pydantic import BaseModel, Field
+from pydantic.generics import GenericModel
+
+T = TypeVar("T", bound=BaseModel)
+
+
+class PaginatedResults(GenericModel, Generic[T]):
+    """Paginated results"""
+    #fmt: off
+    items: list[T] = Field(description="Items")
+    page: int = Field(description="Current Page")
+    pages: int = Field(description="Total number of pages")
+    per_page: int = Field(description="Number of items per page")
+    total: int = Field(description="Total number of items in result")
+    #fmt: on
+
+class ItemStorageABC(ABC, Generic[T]):
+    _on_changed_callbacks: list[Callable[[T], None]]
+    _on_deleted_callbacks: list[Callable[[str], None]]
+
+    def __init__(self) -> None:
+        self._on_changed_callbacks = list()
+        self._on_deleted_callbacks = list()
+
+    """Base item storage class"""
+
+    @abstractmethod
+    def get(self, item_id: str) -> T:
+        pass
+
+    @abstractmethod
+    def set(self, item: T) -> None:
+        pass
+
+    @abstractmethod
+    def list(self, page: int = 0, per_page: int = 10) -> PaginatedResults[T]:
+        pass
+
+    @abstractmethod
+    def search(
+        self, query: str, page: int = 0, per_page: int = 10
+    ) -> PaginatedResults[T]:
+        pass
+
+    def on_changed(self, on_changed: Callable[[T], None]) -> None:
+        """Register a callback for when an item is changed"""
+        self._on_changed_callbacks.append(on_changed)
+
+    def on_deleted(self, on_deleted: Callable[[str], None]) -> None:
+        """Register a callback for when an item is deleted"""
+        self._on_deleted_callbacks.append(on_deleted)
+
+    def _on_changed(self, item: T) -> None:
+        for callback in self._on_changed_callbacks:
+            callback(item)
+
+    def _on_deleted(self, item_id: str) -> None:
+        for callback in self._on_deleted_callbacks:
+            callback(item_id)

invokeai/app/services/model_manager_initializer.py

@@ -0,0 +1,120 @@
+import os
+import sys
+import torch
+from argparse import Namespace
+from invokeai.backend import Args
+from omegaconf import OmegaConf
+from pathlib import Path
+
+import invokeai.version
+from ...backend import ModelManager
+from ...backend.util import choose_precision, choose_torch_device
+from ...backend import Globals
+
+# TODO: Replace with an abstract class base ModelManagerBase
+def get_model_manager(config: Args) -> ModelManager:
+    if not config.conf:
+        config_file = os.path.join(Globals.root, "configs", "models.yaml")
+        if not os.path.exists(config_file):
+            report_model_error(
+                config, FileNotFoundError(f"The file {config_file} could not be found.")
+            )
+
+    print(f">> {invokeai.version.__app_name__}, version {invokeai.version.__version__}")
+    print(f'>> InvokeAI runtime directory is "{Globals.root}"')
+
+    # these two lines prevent a horrible warning message from appearing
+    # when the frozen CLIP tokenizer is imported
+    import transformers  # type: ignore
+
+    transformers.logging.set_verbosity_error()
+    import diffusers
+
+    diffusers.logging.set_verbosity_error()
+
+    # normalize the config directory relative to root
+    if not os.path.isabs(config.conf):
+        config.conf = os.path.normpath(os.path.join(Globals.root, config.conf))
+
+    if config.embeddings:
+        if not os.path.isabs(config.embedding_path):
+            embedding_path = os.path.normpath(
+                os.path.join(Globals.root, config.embedding_path)
+            )
+        else:
+            embedding_path = config.embedding_path
+    else:
+        embedding_path = None
+
+    # migrate legacy models
+    ModelManager.migrate_models()
+
+    # creating the model manager
+    try:
+        device = torch.device(choose_torch_device())
+        precision = 'float16' if config.precision=='float16' \
+        else 'float32' if config.precision=='float32' \
+        else choose_precision(device)
+        
+        model_manager = ModelManager(
+            OmegaConf.load(config.conf),
+            precision=precision,
+            device_type=device,
+            max_loaded_models=config.max_loaded_models,
+            embedding_path = Path(embedding_path),
+        )
+    except (FileNotFoundError, TypeError, AssertionError) as e:
+        report_model_error(config, e)
+    except (IOError, KeyError) as e:
+        print(f"{e}. Aborting.")
+        sys.exit(-1)
+
+    # try to autoconvert new models
+    # autoimport new .ckpt files
+    if path := config.autoconvert:
+        model_manager.autoconvert_weights(
+            conf_path=config.conf,
+            weights_directory=path,
+        )
+
+    return model_manager
+
+def report_model_error(opt: Namespace, e: Exception):
+    print(f'** An error occurred while attempting to initialize the model: "{str(e)}"')
+    print(
+        "** This can be caused by a missing or corrupted models file, and can sometimes be fixed by (re)installing the models."
+    )
+    yes_to_all = os.environ.get("INVOKE_MODEL_RECONFIGURE")
+    if yes_to_all:
+        print(
+            "** Reconfiguration is being forced by environment variable INVOKE_MODEL_RECONFIGURE"
+        )
+    else:
+        response = input(
+            "Do you want to run invokeai-configure script to select and/or reinstall models? [y] "
+        )
+        if response.startswith(("n", "N")):
+            return
+
+    print("invokeai-configure is launching....\n")
+
+    # Match arguments that were set on the CLI
+    # only the arguments accepted by the configuration script are parsed
+    root_dir = ["--root", opt.root_dir] if opt.root_dir is not None else []
+    config = ["--config", opt.conf] if opt.conf is not None else []
+    previous_config = sys.argv
+    sys.argv = ["invokeai-configure"]
+    sys.argv.extend(root_dir)
+    sys.argv.extend(config.to_dict())
+    if yes_to_all is not None:
+        for arg in yes_to_all.split():
+            sys.argv.append(arg)
+
+    from invokeai.frontend.install import invokeai_configure
+
+    invokeai_configure()
+    # TODO: Figure out how to restart
+    # print('** InvokeAI will now restart')
+    # sys.argv = previous_args
+    # main() # would rather do a os.exec(), but doesn't exist?
+    # sys.exit(0)

invokeai/app/services/processor.py

@@ -0,0 +1,121 @@
+import traceback
+from threading import Event, Thread
+
+from ..invocations.baseinvocation import InvocationContext
+from .invocation_queue import InvocationQueueItem
+from .invoker import InvocationProcessorABC, Invoker
+
+
+class DefaultInvocationProcessor(InvocationProcessorABC):
+    __invoker_thread: Thread
+    __stop_event: Event
+    __invoker: Invoker
+
+    def start(self, invoker) -> None:
+        self.__invoker = invoker
+        self.__stop_event = Event()
+        self.__invoker_thread = Thread(
+            name="invoker_processor",
+            target=self.__process,
+            kwargs=dict(stop_event=self.__stop_event),
+        )
+        self.__invoker_thread.daemon = (
+            True  # TODO: probably better to just not use threads?
+        )
+        self.__invoker_thread.start()
+
+    def stop(self, *args, **kwargs) -> None:
+        self.__stop_event.set()
+
+    def __process(self, stop_event: Event):
+        try:
+            while not stop_event.is_set():
+                queue_item: InvocationQueueItem = self.__invoker.services.queue.get()
+                if not queue_item:  # Probably stopping
+                    continue
+
+                graph_execution_state = (
+                    self.__invoker.services.graph_execution_manager.get(
+                        queue_item.graph_execution_state_id
+                    )
+                )
+                invocation = graph_execution_state.execution_graph.get_node(
+                    queue_item.invocation_id
+                )
+
+                # Send starting event
+                self.__invoker.services.events.emit_invocation_started(
+                    graph_execution_state_id=graph_execution_state.id,
+                    invocation_id=invocation.id,
+                )
+
+                # Invoke
+                try:
+                    outputs = invocation.invoke(
+                        InvocationContext(
+                            services=self.__invoker.services,
+                            graph_execution_state_id=graph_execution_state.id,
+                        )
+                    )
+
+                    # Check queue to see if this is canceled, and skip if so
+                    if self.__invoker.services.queue.is_canceled(
+                        graph_execution_state.id
+                    ):
+                        continue
+
+                    # Save outputs and history
+                    graph_execution_state.complete(invocation.id, outputs)
+
+                    # Save the state changes
+                    self.__invoker.services.graph_execution_manager.set(
+                        graph_execution_state
+                    )
+
+                    # Send complete event
+                    self.__invoker.services.events.emit_invocation_complete(
+                        graph_execution_state_id=graph_execution_state.id,
+                        invocation_id=invocation.id,
+                        result=outputs.dict(),
+                    )
+
+                except KeyboardInterrupt:
+                    pass
+
+                except Exception as e:
+                    error = traceback.format_exc()
+
+                    # Save error
+                    graph_execution_state.set_node_error(invocation.id, error)
+
+                    # Save the state changes
+                    self.__invoker.services.graph_execution_manager.set(
+                        graph_execution_state
+                    )
+
+                    # Send error event
+                    self.__invoker.services.events.emit_invocation_error(
+                        graph_execution_state_id=graph_execution_state.id,
+                        invocation_id=invocation.id,
+                        error=error,
+                    )
+
+                    pass
+                
+                # Check queue to see if this is canceled, and skip if so
+                if self.__invoker.services.queue.is_canceled(
+                    graph_execution_state.id
+                ):
+                    continue
+
+                # Queue any further commands if invoking all
+                is_complete = graph_execution_state.is_complete()
+                if queue_item.invoke_all and not is_complete:
+                    self.__invoker.invoke(graph_execution_state, invoke_all=True)
+                elif is_complete:
+                    self.__invoker.services.events.emit_graph_execution_complete(
+                        graph_execution_state.id
+                    )
+
+        except KeyboardInterrupt:
+            ...  # Log something?

invokeai/app/services/restoration_services.py

@@ -0,0 +1,109 @@
+import sys
+import traceback
+import torch
+from ...backend.restoration import Restoration
+from ...backend.util import choose_torch_device, CPU_DEVICE, MPS_DEVICE
+
+# This should be a real base class for postprocessing functions,
+# but right now we just instantiate the existing gfpgan, esrgan
+# and codeformer functions.
+class RestorationServices:
+    '''Face restoration and upscaling'''
+    
+    def __init__(self,args):
+        try:
+            gfpgan, codeformer, esrgan = None, None, None
+            if args.restore or args.esrgan:
+                restoration = Restoration()
+                if args.restore:
+                    gfpgan, codeformer = restoration.load_face_restore_models(
+                        args.gfpgan_model_path
+                    )
+                else:
+                    print(">> Face restoration disabled")
+                    if args.esrgan:
+                        esrgan = restoration.load_esrgan(args.esrgan_bg_tile)
+                    else:
+                        print(">> Upscaling disabled")
+            else:
+                print(">> Face restoration and upscaling disabled")
+        except (ModuleNotFoundError, ImportError):
+            print(traceback.format_exc(), file=sys.stderr)
+            print(">> You may need to install the ESRGAN and/or GFPGAN modules")
+        self.device = torch.device(choose_torch_device())
+        self.gfpgan = gfpgan
+        self.codeformer = codeformer
+        self.esrgan = esrgan
+
+    # note that this one method does gfpgan and codepath reconstruction, as well as
+    # esrgan upscaling
+    # TO DO: refactor into separate methods
+    def upscale_and_reconstruct(
+        self,
+        image_list,
+        facetool="gfpgan",
+        upscale=None,
+        upscale_denoise_str=0.75,
+        strength=0.0,
+        codeformer_fidelity=0.75,
+        save_original=False,
+        image_callback=None,
+        prefix=None,
+    ):
+        results = []
+        for r in image_list:
+            image, seed = r
+            try:
+                if strength > 0:
+                    if self.gfpgan is not None or self.codeformer is not None:
+                        if facetool == "gfpgan":
+                            if self.gfpgan is None:
+                                print(
+                                    ">> GFPGAN not found. Face restoration is disabled."
+                                )
+                            else:
+                                image = self.gfpgan.process(image, strength, seed)
+                        if facetool == "codeformer":
+                            if self.codeformer is None:
+                                print(
+                                    ">> CodeFormer not found. Face restoration is disabled."
+                                )
+                            else:
+                                cf_device = (
+                                    CPU_DEVICE if self.device == MPS_DEVICE else self.device
+                                )
+                                image = self.codeformer.process(
+                                    image=image,
+                                    strength=strength,
+                                    device=cf_device,
+                                    seed=seed,
+                                    fidelity=codeformer_fidelity,
+                                )
+                    else:
+                        print(">> Face Restoration is disabled.")
+                if upscale is not None:
+                    if self.esrgan is not None:
+                        if len(upscale) < 2:
+                            upscale.append(0.75)
+                        image = self.esrgan.process(
+                            image,
+                            upscale[1],
+                            seed,
+                            int(upscale[0]),
+                            denoise_str=upscale_denoise_str,
+                        )
+                    else:
+                        print(">> ESRGAN is disabled. Image not upscaled.")
+            except Exception as e:
+                print(
+                    f">> Error running RealESRGAN or GFPGAN. Your image was not upscaled.\n{e}"
+                )
+
+            if image_callback is not None:
+                image_callback(image, seed, upscaled=True, use_prefix=prefix)
+            else:
+                r[0] = image
+
+            results.append([image, seed])
+
+        return results

invokeai/app/services/sqlite.py

@@ -0,0 +1,138 @@
+import sqlite3
+from threading import Lock
+from typing import Generic, TypeVar, Union, get_args
+
+from pydantic import BaseModel, parse_raw_as
+
+from .item_storage import ItemStorageABC, PaginatedResults
+
+T = TypeVar("T", bound=BaseModel)
+
+sqlite_memory = ":memory:"
+
+
+class SqliteItemStorage(ItemStorageABC, Generic[T]):
+    _filename: str
+    _table_name: str
+    _conn: sqlite3.Connection
+    _cursor: sqlite3.Cursor
+    _id_field: str
+    _lock: Lock
+
+    def __init__(self, filename: str, table_name: str, id_field: str = "id"):
+        super().__init__()
+
+        self._filename = filename
+        self._table_name = table_name
+        self._id_field = id_field  # TODO: validate that T has this field
+        self._lock = Lock()
+
+        self._conn = sqlite3.connect(
+            self._filename, check_same_thread=False
+        )  # TODO: figure out a better threading solution
+        self._cursor = self._conn.cursor()
+
+        self._create_table()
+
+    def _create_table(self):
+        try:
+            self._lock.acquire()
+            self._cursor.execute(
+                f"""CREATE TABLE IF NOT EXISTS {self._table_name} (
+                item TEXT,
+                id TEXT GENERATED ALWAYS AS (json_extract(item, '$.{self._id_field}')) VIRTUAL NOT NULL);"""
+            )
+            self._cursor.execute(
+                f"""CREATE UNIQUE INDEX IF NOT EXISTS {self._table_name}_id ON {self._table_name}(id);"""
+            )
+        finally:
+            self._lock.release()
+
+    def _parse_item(self, item: str) -> T:
+        item_type = get_args(self.__orig_class__)[0]
+        return parse_raw_as(item_type, item)
+
+    def set(self, item: T):
+        try:
+            self._lock.acquire()
+            self._cursor.execute(
+                f"""INSERT OR REPLACE INTO {self._table_name} (item) VALUES (?);""",
+                (item.json(),),
+            )
+        finally:
+            self._lock.release()
+        self._on_changed(item)
+
+    def get(self, id: str) -> Union[T, None]:
+        try:
+            self._lock.acquire()
+            self._cursor.execute(
+                f"""SELECT item FROM {self._table_name} WHERE id = ?;""", (str(id),)
+            )
+            result = self._cursor.fetchone()
+        finally:
+            self._lock.release()
+
+        if not result:
+            return None
+
+        return self._parse_item(result[0])
+
+    def delete(self, id: str):
+        try:
+            self._lock.acquire()
+            self._cursor.execute(
+                f"""DELETE FROM {self._table_name} WHERE id = ?;""", (str(id),)
+            )
+        finally:
+            self._lock.release()
+        self._on_deleted(id)
+
+    def list(self, page: int = 0, per_page: int = 10) -> PaginatedResults[T]:
+        try:
+            self._lock.acquire()
+            self._cursor.execute(
+                f"""SELECT item FROM {self._table_name} LIMIT ? OFFSET ?;""",
+                (per_page, page * per_page),
+            )
+            result = self._cursor.fetchall()
+
+            items = list(map(lambda r: self._parse_item(r[0]), result))
+
+            self._cursor.execute(f"""SELECT count(*) FROM {self._table_name};""")
+            count = self._cursor.fetchone()[0]
+        finally:
+            self._lock.release()
+
+        pageCount = int(count / per_page) + 1
+
+        return PaginatedResults[T](
+            items=items, page=page, pages=pageCount, per_page=per_page, total=count
+        )
+
+    def search(
+        self, query: str, page: int = 0, per_page: int = 10
+    ) -> PaginatedResults[T]:
+        try:
+            self._lock.acquire()
+            self._cursor.execute(
+                f"""SELECT item FROM {self._table_name} WHERE item LIKE ? LIMIT ? OFFSET ?;""",
+                (f"%{query}%", per_page, page * per_page),
+            )
+            result = self._cursor.fetchall()
+
+            items = list(map(lambda r: self._parse_item(r[0]), result))
+
+            self._cursor.execute(
+                f"""SELECT count(*) FROM {self._table_name} WHERE item LIKE ?;""",
+                (f"%{query}%",),
+            )
+            count = self._cursor.fetchone()[0]
+        finally:
+            self._lock.release()
+
+        pageCount = int(count / per_page) + 1
+
+        return PaginatedResults[T](
+            items=items, page=page, pages=pageCount, per_page=per_page, total=count
+        )

invokeai/backend/__init__.py

@@ -1,5 +1,16 @@
-'''
+"""
 Initialization file for invokeai.backend
-'''
-from .invoke_ai_web_server import InvokeAIWebServer
-
+"""
+from .generate import Generate
+from .generator import (
+    InvokeAIGeneratorBasicParams,
+    InvokeAIGenerator,
+    InvokeAIGeneratorOutput,
+    Txt2Img,
+    Img2Img,
+    Inpaint
+)
+from .model_management import ModelManager
+from .safety_checker import SafetyChecker
+from .args import Args
+from .globals import Globals

invokeai/backend/config/invokeai_configure.py

@@ -6,19 +6,20 @@
 #
 # Coauthor: Kevin Turner http://github.com/keturn
 #
-print("Loading Python libraries...\n")
+import sys
+print("Loading Python libraries...\n",file=sys.stderr)
+
 import argparse
 import io
 import os
 import re
 import shutil
-import sys
 import traceback
 import warnings
 from argparse import Namespace
 from pathlib import Path
-from urllib import request
 from shutil import get_terminal_size
+from urllib import request
 
 import npyscreen
 import torch
@@ -37,17 +38,20 @@ from transformers import (
 
 import invokeai.configs as configs
 
+from ...frontend.install.model_install import addModelsForm, process_and_execute
+from ...frontend.install.widgets import (
+    CenteredButtonPress,
+    IntTitleSlider,
+    set_min_terminal_size,
+)
 from ..args import PRECISION_CHOICES, Args
-from ..globals import Globals, global_config_dir, global_config_file, global_cache_dir
-from .model_install import addModelsForm, process_and_execute
+from ..globals import Globals, global_cache_dir, global_config_dir, global_config_file
 from .model_install_backend import (
     default_dataset,
     download_from_hf,
-    recommended_datasets,
     hf_download_with_resume,
+    recommended_datasets,
 )
-from .widgets import IntTitleSlider, CenteredButtonPress, set_min_terminal_size
-
 
 warnings.filterwarnings("ignore")
 
@@ -82,6 +86,7 @@ INIT_FILE_PREAMBLE = """# InvokeAI initialization file
 # -Ak_euler_a -C10.0
 """
 
+
 # --------------------------------------------
 def postscript(errors: None):
     if not any(errors):
@@ -180,13 +185,11 @@ def download_with_progress_bar(model_url: str, model_dest: str, label: str = "th
 # ---------------------------------------------
 # this will preload the Bert tokenizer fles
 def download_bert():
-    print(
-        "Installing bert tokenizer...",
-        file=sys.stderr
-    )
+    print("Installing bert tokenizer...", file=sys.stderr)
     with warnings.catch_warnings():
         warnings.filterwarnings("ignore", category=DeprecationWarning)
         from transformers import BertTokenizerFast
+
         download_from_hf(BertTokenizerFast, "bert-base-uncased")
 
 
@@ -197,12 +200,14 @@ def download_sd1_clip():
     download_from_hf(CLIPTokenizer, version)
     download_from_hf(CLIPTextModel, version)
 
+
 # ---------------------------------------------
 def download_sd2_clip():
-    version = 'stabilityai/stable-diffusion-2'
+    version = "stabilityai/stable-diffusion-2"
     print("Installing SD2 clip model...", file=sys.stderr)
-    download_from_hf(CLIPTokenizer, version, subfolder='tokenizer')
-    download_from_hf(CLIPTextModel, version, subfolder='text_encoder')
+    download_from_hf(CLIPTokenizer, version, subfolder="tokenizer")
+    download_from_hf(CLIPTextModel, version, subfolder="text_encoder")
+
 
 # ---------------------------------------------
 def download_realesrgan():
@@ -323,13 +328,13 @@ def get_root(root: str = None) -> str:
 class editOptsForm(npyscreen.FormMultiPage):
     # for responsive resizing - disabled
     # FIX_MINIMUM_SIZE_WHEN_CREATED = False
-    
+
     def create(self):
         program_opts = self.parentApp.program_opts
         old_opts = self.parentApp.invokeai_opts
         first_time = not (Globals.root / Globals.initfile).exists()
         access_token = HfFolder.get_token()
-        window_width,window_height = get_terminal_size()
+        window_width, window_height = get_terminal_size()
         for i in [
             "Configure startup settings. You can come back and change these later.",
             "Use ctrl-N and ctrl-P to move to the <N>ext and <P>revious fields.",
@@ -681,6 +686,7 @@ def run_console_ui(
     else:
         return (editApp.new_opts, editApp.user_selections)
 
+
 # -------------------------------------
 def write_opts(opts: Namespace, init_file: Path):
     """
@@ -701,8 +707,8 @@ def write_opts(opts: Namespace, init_file: Path):
         "^--?(o|out|no-xformer|xformer|no-ckpt|ckpt|free|no-nsfw|nsfw|prec|max_load|embed|always|ckpt|free_gpu)"
     )
     # fix windows paths
-    opts.outdir = opts.outdir.replace('\\','/')
-    opts.embedding_path = opts.embedding_path.replace('\\','/')
+    opts.outdir = opts.outdir.replace("\\", "/")
+    opts.embedding_path = opts.embedding_path.replace("\\", "/")
     new_file = f"{init_file}.new"
     try:
         lines = [x.strip() for x in open(init_file, "r").readlines()]
@@ -712,8 +718,8 @@ def write_opts(opts: Namespace, init_file: Path):
                     out_file.write(line + "\n")
             out_file.write(
                 f"""
---outdir="{opts.outdir}"
---embedding_path="{opts.embedding_path}"
+--outdir={opts.outdir}
+--embedding_path={opts.embedding_path}
 --precision={opts.precision}
 --max_loaded_models={int(opts.max_loaded_models)}
 --{'no-' if not opts.safety_checker else ''}nsfw_checker
@@ -855,6 +861,7 @@ def main():
     except KeyboardInterrupt:
         print("\nGoodbye! Come back soon.")
 
+
 # -------------------------------------
 if __name__ == "__main__":
     main()

invokeai/backend/config/model_install_backend.py

@@ -8,6 +8,7 @@ import sys
 import warnings
 from pathlib import Path
 from tempfile import TemporaryFile
+from typing import List
 
 import requests
 from diffusers import AutoencoderKL
@@ -15,12 +16,12 @@ from huggingface_hub import hf_hub_url
 from omegaconf import OmegaConf
 from omegaconf.dictconfig import DictConfig
 from tqdm import tqdm
-from typing import List
 
 import invokeai.configs as configs
-from ..generator.diffusers_pipeline import StableDiffusionGeneratorPipeline
+
 from ..globals import Globals, global_cache_dir, global_config_dir
-from ..model_manager import ModelManager
+from ..model_management import ModelManager
+from ..stable_diffusion import StableDiffusionGeneratorPipeline
 
 warnings.filterwarnings("ignore")
 
@@ -44,45 +45,49 @@ Config_preamble = """
 # was trained on.
 """
 
+
 def default_config_file():
     return Path(global_config_dir()) / "models.yaml"
 
+
 def sd_configs():
     return Path(global_config_dir()) / "stable-diffusion"
 
+
 def initial_models():
     global Datasets
     if Datasets:
         return Datasets
     return (Datasets := OmegaConf.load(Dataset_path))
 
+
 def install_requested_models(
-        install_initial_models: List[str] = None,
-        remove_models: List[str] = None,
-        scan_directory: Path = None,
-        external_models: List[str] = None,
-        scan_at_startup: bool = False,
-        convert_to_diffusers: bool = False,
-        precision: str = "float16",
-        purge_deleted: bool = False,
-        config_file_path: Path = None,
+    install_initial_models: List[str] = None,
+    remove_models: List[str] = None,
+    scan_directory: Path = None,
+    external_models: List[str] = None,
+    scan_at_startup: bool = False,
+    convert_to_diffusers: bool = False,
+    precision: str = "float16",
+    purge_deleted: bool = False,
+    config_file_path: Path = None,
 ):
-    '''
+    """
     Entry point for installing/deleting starter models, or installing external models.
-    '''
-    config_file_path=config_file_path or default_config_file()
+    """
+    config_file_path = config_file_path or default_config_file()
     if not config_file_path.exists():
-        open(config_file_path,'w')
-            
-    model_manager= ModelManager(OmegaConf.load(config_file_path),precision=precision)
-    
+        open(config_file_path, "w")
+
+    model_manager = ModelManager(OmegaConf.load(config_file_path), precision=precision)
+
     if remove_models and len(remove_models) > 0:
         print("== DELETING UNCHECKED STARTER MODELS ==")
         for model in remove_models:
-            print(f'{model}...')
+            print(f"{model}...")
             model_manager.del_model(model, delete_files=purge_deleted)
         model_manager.commit(config_file_path)
-    
+
     if install_initial_models and len(install_initial_models) > 0:
         print("== INSTALLING SELECTED STARTER MODELS ==")
         successfully_downloaded = download_weight_datasets(
@@ -96,21 +101,20 @@ def install_requested_models(
 
     # due to above, we have to reload the model manager because conf file
     # was changed behind its back
-    model_manager= ModelManager(OmegaConf.load(config_file_path),precision=precision)
+    model_manager = ModelManager(OmegaConf.load(config_file_path), precision=precision)
 
     external_models = external_models or list()
     if scan_directory:
         external_models.append(str(scan_directory))
 
-    if len(external_models)>0:
+    if len(external_models) > 0:
         print("== INSTALLING EXTERNAL MODELS ==")
         for path_url_or_repo in external_models:
             try:
                 model_manager.heuristic_import(
                     path_url_or_repo,
                     convert=convert_to_diffusers,
-                    config_file_callback=_pick_configuration_file,
-                    commit_to_conf=config_file_path
+                    commit_to_conf=config_file_path,
                 )
             except KeyboardInterrupt:
                 sys.exit(-1)
@@ -118,17 +122,18 @@ def install_requested_models(
                 pass
 
     if scan_at_startup and scan_directory.is_dir():
-        argument = '--autoconvert' if convert_to_diffusers else '--autoimport'
+        argument = "--autoconvert" if convert_to_diffusers else "--autoimport"
         initfile = Path(Globals.root, Globals.initfile)
-        replacement = Path(Globals.root, f'{Globals.initfile}.new')
-        directory = str(scan_directory).replace('\\','/')
-        with open(initfile,'r') as input:
-            with open(replacement,'w') as output:
+        replacement = Path(Globals.root, f"{Globals.initfile}.new")
+        directory = str(scan_directory).replace("\\", "/")
+        with open(initfile, "r") as input:
+            with open(replacement, "w") as output:
                 while line := input.readline():
                     if not line.startswith(argument):
                         output.writelines([line])
-                output.writelines([f'{argument} "{directory}"'])
-        os.replace(replacement,initfile)
+                output.writelines([f"{argument} {directory}"])
+        os.replace(replacement, initfile)
+
 
 # -------------------------------------
 def yes_or_no(prompt: str, default_yes=True):
@@ -139,45 +144,6 @@ def yes_or_no(prompt: str, default_yes=True):
     else:
         return response[0] in ("y", "Y")
 
-# -------------------------------------
-def _pick_configuration_file(checkpoint_path: Path)->Path:
-    print(
-"""
-Please select the type of this model:
-[1] A Stable Diffusion v1.x ckpt/safetensors model
-[2] A Stable Diffusion v1.x inpainting ckpt/safetensors model
-[3] A Stable Diffusion v2.x base model (512 pixels; no 'parameterization:' in its yaml file)
-[4] A Stable Diffusion v2.x v-predictive model (768 pixels; look for 'parameterization: "v"' in its yaml file)
-[5] Other (you will be prompted to enter the config file path)
-[Q] I have no idea! Skip the import.
-""")
-    choices = [
-        global_config_dir() / 'stable-diffusion' / x
-        for x in [
-                'v1-inference.yaml',
-                'v1-inpainting-inference.yaml',
-                'v2-inference.yaml',
-                'v2-inference-v.yaml',
-        ]
-    ]
-
-    ok = False
-    while not ok:
-        try:
-            choice = input('select 0-5, Q > ').strip()
-            if choice.startswith(('q','Q')):
-                return
-            if choice == '5':
-                choice = Path(input('Select config file for this model> ').strip()).absolute()
-                ok = choice.exists()
-            else:
-                choice = choices[int(choice)-1]
-                ok = True
-        except (ValueError, IndexError):
-            print(f'{choice} is not a valid choice')
-        except EOFError:
-            return
-    return choice
 
 # -------------------------------------
 def get_root(root: str = None) -> str:
@@ -223,7 +189,9 @@ def migrate_models_ckpt():
     if not os.path.exists(os.path.join(model_path, "model.ckpt")):
         return
     new_name = initial_models()["stable-diffusion-1.4"]["file"]
-    print('The Stable Diffusion v4.1 "model.ckpt" is already installed. The name will be changed to {new_name} to avoid confusion.')
+    print(
+        'The Stable Diffusion v4.1 "model.ckpt" is already installed. The name will be changed to {new_name} to avoid confusion.'
+    )
     print(f"model.ckpt => {new_name}")
     os.replace(
         os.path.join(model_path, "model.ckpt"), os.path.join(model_path, new_name)
@@ -422,7 +390,8 @@ def update_config_file(successfully_downloaded: dict, config_file: Path):
 
 # ---------------------------------------------
 def new_config_file_contents(
-        successfully_downloaded: dict, config_file: Path,
+    successfully_downloaded: dict,
+    config_file: Path,
 ) -> str:
     if config_file.exists():
         conf = OmegaConf.load(str(config_file.expanduser().resolve()))
@@ -452,7 +421,9 @@ def new_config_file_contents(
             stanza["weights"] = os.path.relpath(
                 successfully_downloaded[model], start=Globals.root
             )
-            stanza["config"] = os.path.normpath(os.path.join(sd_configs(), mod["config"]))
+            stanza["config"] = os.path.normpath(
+                os.path.join(sd_configs(), mod["config"])
+            )
         if "vae" in mod:
             if "file" in mod["vae"]:
                 stanza["vae"] = os.path.normpath(
@@ -484,7 +455,7 @@ def delete_weights(model_name: str, conf_stanza: dict):
     print(
         f"\n** The checkpoint version of {model_name} is superseded by the diffusers version. Deleting the original file {weights}?"
     )
-          
+
     weights = Path(weights)
     if not weights.is_absolute():
         weights = Path(Globals.root) / weights

invokeai/backend/generate.py

@@ -5,6 +5,7 @@
 
 import gc
 import importlib
+import logging
 import os
 import random
 import re
@@ -19,28 +20,23 @@ import numpy as np
 import skimage
 import torch
 import transformers
+from PIL import Image, ImageOps
+from accelerate.utils import set_seed
 from diffusers.pipeline_utils import DiffusionPipeline
 from diffusers.utils.import_utils import is_xformers_available
 from omegaconf import OmegaConf
-from PIL import Image, ImageOps
-from pytorch_lightning import logging, seed_everything
-
-import ldm.invoke.conditioning
-from ldm.invoke.args import metadata_from_png
-from ldm.invoke.concepts_lib import HuggingFaceConceptsLibrary
-from ldm.invoke.conditioning import get_uc_and_c_and_ec
-from ldm.invoke.devices import choose_precision, choose_torch_device
-from ldm.invoke.generator.inpaint import infill_methods
-from ldm.invoke.globals import Globals, global_cache_dir
-from ldm.invoke.image_util import InitImageResizer
-from ldm.invoke.model_manager import ModelManager
-from ldm.invoke.pngwriter import PngWriter
-from ldm.invoke.seamless import configure_model_padding
-from ldm.invoke.txt2mask import Txt2Mask
-from ldm.models.diffusion.ddim import DDIMSampler
-from ldm.models.diffusion.ksampler import KSampler
-from ldm.models.diffusion.plms import PLMSSampler
+from pathlib import Path
 
+from .args import metadata_from_png
+from .generator import infill_methods
+from .globals import Globals, global_cache_dir
+from .image_util import InitImageResizer, PngWriter, Txt2Mask, configure_model_padding
+from .model_management import ModelManager
+from .safety_checker import SafetyChecker
+from .prompting import get_uc_and_c_and_ec
+from .prompting.conditioning import log_tokenization
+from .stable_diffusion import HuggingFaceConceptsLibrary
+from .util import choose_precision, choose_torch_device
 
 def fix_func(orig):
     if hasattr(torch.backends, "mps") and torch.backends.mps.is_available():
@@ -200,8 +196,6 @@ class Generate:
         # it wasn't actually doing anything. This logic could be reinstated.
         self.device = torch.device(choose_torch_device())
         print(f">> Using device_type {self.device.type}")
-        if self.device.type == 'cuda':
-            print(f">> CUDA device '{torch.cuda.get_device_name(torch.cuda.current_device())}' (GPU {os.environ.get('CUDA_VISIBLE_DEVICES') or 0})")
         if full_precision:
             if self.precision != "auto":
                 raise ValueError("Remove --full_precision / -F if using --precision")
@@ -229,6 +223,7 @@ class Generate:
             self.precision,
             max_loaded_models=max_loaded_models,
             sequential_offload=self.free_gpu_mem,
+            embedding_path=Path(self.embedding_path),
         )
         # don't accept invalid models
         fallback = self.model_manager.default_model() or FALLBACK_MODEL_NAME
@@ -251,31 +246,8 @@ class Generate:
 
         # load safety checker if requested
         if safety_checker:
-            try:
-                print(">> Initializing NSFW checker")
-                from diffusers.pipelines.stable_diffusion.safety_checker import (
-                    StableDiffusionSafetyChecker,
-                )
-                from transformers import AutoFeatureExtractor
-
-                safety_model_id = "CompVis/stable-diffusion-safety-checker"
-                safety_model_path = global_cache_dir("hub")
-                self.safety_checker = StableDiffusionSafetyChecker.from_pretrained(
-                    safety_model_id,
-                    local_files_only=True,
-                    cache_dir=safety_model_path,
-                )
-                self.safety_feature_extractor = AutoFeatureExtractor.from_pretrained(
-                    safety_model_id,
-                    local_files_only=True,
-                    cache_dir=safety_model_path,
-                )
-                self.safety_checker.to(self.device)
-            except Exception:
-                print(
-                    "** An error was encountered while installing the safety checker:"
-                )
-                print(traceback.format_exc())
+            print(">> Initializing NSFW checker")
+            self.safety_checker = SafetyChecker(self.device)
         else:
             print(">> NSFW checker is disabled")
 
@@ -330,8 +302,8 @@ class Generate:
         variation_amount=0.0,
         threshold=0.0,
         perlin=0.0,
-        h_symmetry_time_pct = None,
-        v_symmetry_time_pct = None,
+        h_symmetry_time_pct=None,
+        v_symmetry_time_pct=None,
         karras_max=None,
         outdir=None,
         # these are specific to img2img and inpaint
@@ -488,7 +460,7 @@ class Generate:
 
         if sampler_name and (sampler_name != self.sampler_name):
             self.sampler_name = sampler_name
-            self._set_sampler()
+            self._set_scheduler()
 
         # apply the concepts library to the prompt
         prompt = self.huggingface_concepts_library.replace_concepts_with_triggers(
@@ -497,28 +469,11 @@ class Generate:
             self.model.textual_inversion_manager.get_all_trigger_strings(),
         )
 
-        # bit of a hack to change the cached sampler's karras threshold to
-        # whatever the user asked for
-        if karras_max is not None and isinstance(self.sampler, KSampler):
-            self.sampler.adjust_settings(karras_max=karras_max)
-
         tic = time.time()
         if self._has_cuda():
             torch.cuda.reset_peak_memory_stats()
 
         results = list()
-        init_image = None
-        mask_image = None
-
-        try:
-            if (
-                self.free_gpu_mem
-                and self.model.cond_stage_model.device != self.model.device
-            ):
-                self.model.cond_stage_model.device = self.model.device
-                self.model.cond_stage_model.to(self.model.device)
-        except AttributeError:
-            pass
 
         try:
             uc, c, extra_conditioning_info = get_uc_and_c_and_ec(
@@ -547,15 +502,6 @@ class Generate:
             generator.set_variation(self.seed, variation_amount, with_variations)
             generator.use_mps_noise = use_mps_noise
 
-            checker = (
-                {
-                    "checker": self.safety_checker,
-                    "extractor": self.safety_feature_extractor,
-                }
-                if self.safety_checker
-                else None
-            )
-
             results = generator.generate(
                 prompt,
                 iterations=iterations,
@@ -582,7 +528,7 @@ class Generate:
                 embiggen_strength=embiggen_strength,
                 inpaint_replace=inpaint_replace,
                 mask_blur_radius=mask_blur_radius,
-                safety_checker=checker,
+                safety_checker=self.safety_checker,
                 seam_size=seam_size,
                 seam_blur=seam_blur,
                 seam_strength=seam_strength,
@@ -719,7 +665,7 @@ class Generate:
             prompt,
             model=self.model,
             skip_normalize_legacy_blend=opt.skip_normalize,
-            log_tokens=ldm.invoke.conditioning.log_tokenization,
+            log_tokens=log_tokenization,
         )
 
         if tool in ("gfpgan", "codeformer", "upscale"):
@@ -743,7 +689,7 @@ class Generate:
             )
 
         elif tool == "outcrop":
-            from ldm.invoke.restoration.outcrop import Outcrop
+            from .restoration.outcrop import Outcrop
 
             extend_instructions = {}
             for direction, pixels in _pairwise(opt.outcrop):
@@ -796,7 +742,7 @@ class Generate:
                 clear_cuda_cache=self.clear_cuda_cache,
             )
         elif tool == "outpaint":
-            from ldm.invoke.restoration.outpaint import Outpaint
+            from .restoration.outpaint import Outpaint
 
             restorer = Outpaint(image, self)
             return restorer.process(opt, args, image_callback=callback, prefix=prefix)
@@ -818,17 +764,12 @@ class Generate:
         hires_fix: bool = False,
         force_outpaint: bool = False,
     ):
-        inpainting_model_in_use = self.sampler.uses_inpainting_model()
-
         if hires_fix:
             return self._make_txt2img2img()
 
         if embiggen is not None:
             return self._make_embiggen()
 
-        if inpainting_model_in_use:
-            return self._make_omnibus()
-
         if ((init_image is not None) and (mask_image is not None)) or force_outpaint:
             return self._make_inpaint()
 
@@ -905,16 +846,9 @@ class Generate:
     def _make_inpaint(self):
         return self._load_generator(".inpaint", "Inpaint")
 
-    def _make_omnibus(self):
-        return self._load_generator(".omnibus", "Omnibus")
-
     def _load_generator(self, module, class_name):
-        if self.is_legacy_model(self.model_name):
-            mn = f"ldm.invoke.ckpt_generator{module}"
-            cn = f"Ckpt{class_name}"
-        else:
-            mn = f"ldm.invoke.generator{module}"
-            cn = class_name
+        mn = f"invokeai.backend.generator{module}"
+        cn = class_name
         module = importlib.import_module(mn)
         constructor = getattr(module, cn)
         return constructor(self.model, self.precision)
@@ -975,21 +909,9 @@ class Generate:
         # uncache generators so they pick up new models
         self.generators = {}
 
-        seed_everything(random.randrange(0, np.iinfo(np.uint32).max))
-        if self.embedding_path is not None:
-            print(f'>> Loading embeddings from {self.embedding_path}')
-            for root, _, files in os.walk(self.embedding_path):
-                for name in files:
-                    ti_path = os.path.join(root, name)
-                    self.model.textual_inversion_manager.load_textual_inversion(
-                        ti_path, defer_injecting_tokens=True
-                    )
-            print(
-                f'>> Textual inversion triggers: {", ".join(sorted(self.model.textual_inversion_manager.get_all_trigger_strings()))}'
-            )
-
+        set_seed(random.randrange(0, np.iinfo(np.uint32).max))
         self.model_name = model_name
-        self._set_sampler()  # requires self.model_name to be set first
+        self._set_scheduler()  # requires self.model_name to be set first
         return self.model
 
     def load_huggingface_concepts(self, concepts: list[str]):
@@ -1032,8 +954,9 @@ class Generate:
         image_callback=None,
         prefix=None,
     ):
+        results = []
         for r in image_list:
-            image, seed = r
+            image, seed, _ = r
             try:
                 if strength > 0:
                     if self.gfpgan is not None or self.codeformer is not None:
@@ -1085,6 +1008,10 @@ class Generate:
             else:
                 r[0] = image
 
+            results.append([image, seed])
+
+        return results
+
     def apply_textmask(
         self, image_path: str, prompt: str, callback, threshold: float = 0.5
     ):
@@ -1116,44 +1043,6 @@ class Generate:
     def is_legacy_model(self, model_name) -> bool:
         return self.model_manager.is_legacy(model_name)
 
-    def _set_sampler(self):
-        if isinstance(self.model, DiffusionPipeline):
-            return self._set_scheduler()
-        else:
-            return self._set_sampler_legacy()
-
-    # very repetitive code - can this be simplified? The KSampler names are
-    # consistent, at least
-    def _set_sampler_legacy(self):
-        msg = f">> Setting Sampler to {self.sampler_name}"
-        if self.sampler_name == "plms":
-            self.sampler = PLMSSampler(self.model, device=self.device)
-        elif self.sampler_name == "ddim":
-            self.sampler = DDIMSampler(self.model, device=self.device)
-        elif self.sampler_name == "k_dpm_2_a":
-            self.sampler = KSampler(self.model, "dpm_2_ancestral", device=self.device)
-        elif self.sampler_name == "k_dpm_2":
-            self.sampler = KSampler(self.model, "dpm_2", device=self.device)
-        elif self.sampler_name == "k_dpmpp_2_a":
-            self.sampler = KSampler(
-                self.model, "dpmpp_2s_ancestral", device=self.device
-            )
-        elif self.sampler_name == "k_dpmpp_2":
-            self.sampler = KSampler(self.model, "dpmpp_2m", device=self.device)
-        elif self.sampler_name == "k_euler_a":
-            self.sampler = KSampler(self.model, "euler_ancestral", device=self.device)
-        elif self.sampler_name == "k_euler":
-            self.sampler = KSampler(self.model, "euler", device=self.device)
-        elif self.sampler_name == "k_heun":
-            self.sampler = KSampler(self.model, "heun", device=self.device)
-        elif self.sampler_name == "k_lms":
-            self.sampler = KSampler(self.model, "lms", device=self.device)
-        else:
-            msg = f">> Unsupported Sampler: {self.sampler_name}, Defaulting to plms"
-            self.sampler = PLMSSampler(self.model, device=self.device)
-
-        print(msg)
-
     def _set_scheduler(self):
         default = self.model.scheduler
 

invokeai/backend/generator/__init__.py

@@ -0,0 +1,13 @@
+"""
+Initialization file for the invokeai.generator package
+"""
+from .base import (
+    InvokeAIGenerator,
+    InvokeAIGeneratorBasicParams,
+    InvokeAIGeneratorOutput,
+    Txt2Img,
+    Img2Img,
+    Inpaint,
+    Generator,
+)
+from .inpaint import infill_methods

invokeai/backend/generator/base.py

@@ -0,0 +1,648 @@
+"""
+Base class for invokeai.backend.generator.*
+including img2img, txt2img, and inpaint
+"""
+from __future__ import annotations
+
+import itertools
+import dataclasses
+import diffusers
+import os
+import random
+import traceback
+from abc import ABCMeta
+from argparse import Namespace
+from contextlib import nullcontext
+
+import cv2
+import numpy as np
+import torch
+from PIL import Image, ImageChops, ImageFilter
+from accelerate.utils import set_seed
+from diffusers import DiffusionPipeline
+from tqdm import trange
+from typing import List, Iterator, Type
+from dataclasses import dataclass, field
+from diffusers.schedulers import SchedulerMixin as Scheduler
+
+from ..image_util import configure_model_padding
+from ..util.util import rand_perlin_2d
+from ..safety_checker import SafetyChecker
+from ..prompting.conditioning import get_uc_and_c_and_ec
+from ..stable_diffusion.diffusers_pipeline import StableDiffusionGeneratorPipeline
+
+downsampling = 8
+
+@dataclass
+class InvokeAIGeneratorBasicParams:
+    seed: int=None
+    width: int=512
+    height: int=512
+    cfg_scale: int=7.5
+    steps: int=20
+    ddim_eta: float=0.0
+    scheduler: int='ddim'
+    precision: str='float16'
+    perlin: float=0.0
+    threshold: int=0.0
+    seamless: bool=False
+    seamless_axes: List[str]=field(default_factory=lambda: ['x', 'y'])
+    h_symmetry_time_pct: float=None
+    v_symmetry_time_pct: float=None
+    variation_amount: float = 0.0
+    with_variations: list=field(default_factory=list)
+    safety_checker: SafetyChecker=None
+
+@dataclass
+class InvokeAIGeneratorOutput:
+    '''
+    InvokeAIGeneratorOutput is a dataclass that contains the outputs of a generation
+    operation, including the image, its seed, the model name used to generate the image
+    and the model hash, as well as all the generate() parameters that went into
+    generating the image (in .params, also available as attributes)
+    '''
+    image: Image
+    seed: int
+    model_hash: str
+    attention_maps_images: List[Image]
+    params: Namespace
+
+# we are interposing a wrapper around the original Generator classes so that
+# old code that calls Generate will continue to work.
+class InvokeAIGenerator(metaclass=ABCMeta):
+    scheduler_map = dict(
+        ddim=diffusers.DDIMScheduler,
+        dpmpp_2=diffusers.DPMSolverMultistepScheduler,
+        k_dpm_2=diffusers.KDPM2DiscreteScheduler,
+        k_dpm_2_a=diffusers.KDPM2AncestralDiscreteScheduler,
+        k_dpmpp_2=diffusers.DPMSolverMultistepScheduler,
+        k_euler=diffusers.EulerDiscreteScheduler,
+        k_euler_a=diffusers.EulerAncestralDiscreteScheduler,
+        k_heun=diffusers.HeunDiscreteScheduler,
+        k_lms=diffusers.LMSDiscreteScheduler,
+        plms=diffusers.PNDMScheduler,
+    )
+
+    def __init__(self,
+                 model_info: dict,
+                 params: InvokeAIGeneratorBasicParams=InvokeAIGeneratorBasicParams(),
+                 ):
+        self.model_info=model_info
+        self.params=params
+
+    def generate(self,
+                 prompt: str='',
+                 callback: callable=None,
+                 step_callback: callable=None,
+                 iterations: int=1,
+                 **keyword_args,
+                 )->Iterator[InvokeAIGeneratorOutput]:
+        '''
+        Return an iterator across the indicated number of generations.
+        Each time the iterator is called it will return an InvokeAIGeneratorOutput
+        object. Use like this:
+
+           outputs = txt2img.generate(prompt='banana sushi', iterations=5)
+           for result in outputs:
+               print(result.image, result.seed)
+
+        In the typical case of wanting to get just a single image, iterations
+        defaults to 1 and do:
+
+           output = next(txt2img.generate(prompt='banana sushi')
+
+        Pass None to get an infinite iterator.
+
+           outputs = txt2img.generate(prompt='banana sushi', iterations=None)
+           for o in outputs:
+               print(o.image, o.seed)
+
+        '''
+        generator_args = dataclasses.asdict(self.params)
+        generator_args.update(keyword_args)
+
+        model_info = self.model_info
+        model_name = model_info['model_name']
+        model:StableDiffusionGeneratorPipeline = model_info['model']
+        model_hash = model_info['hash']
+        scheduler: Scheduler = self.get_scheduler(
+            model=model,
+            scheduler_name=generator_args.get('scheduler')
+        )
+        uc, c, extra_conditioning_info = get_uc_and_c_and_ec(prompt,model=model)
+        gen_class = self._generator_class()
+        generator = gen_class(model, self.params.precision)
+        if self.params.variation_amount > 0:
+            generator.set_variation(generator_args.get('seed'),
+                                    generator_args.get('variation_amount'),
+                                    generator_args.get('with_variations')
+                                    )
+
+        if isinstance(model, DiffusionPipeline):
+            for component in [model.unet, model.vae]:
+                configure_model_padding(component,
+                                        generator_args.get('seamless',False),
+                                        generator_args.get('seamless_axes')
+                                        )
+        else:
+            configure_model_padding(model,
+                                    generator_args.get('seamless',False),
+                                    generator_args.get('seamless_axes')
+                                    )
+
+        iteration_count = range(iterations) if iterations else itertools.count(start=0, step=1)
+        for i in iteration_count:
+            results = generator.generate(prompt,
+                                         conditioning=(uc, c, extra_conditioning_info),
+                                         sampler=scheduler,
+                                         **generator_args,
+                                         )
+            output = InvokeAIGeneratorOutput(
+                image=results[0][0],
+                seed=results[0][1],
+                attention_maps_images=results[0][2],
+                model_hash = model_hash,
+                params=Namespace(model_name=model_name,**generator_args),
+            )
+            if callback:
+                callback(output)
+            yield output
+
+    @classmethod
+    def schedulers(self)->List[str]:
+        '''
+        Return list of all the schedulers that we currently handle.
+        '''
+        return list(self.scheduler_map.keys())
+
+    def load_generator(self, model: StableDiffusionGeneratorPipeline, generator_class: Type[Generator]):
+        return generator_class(model, self.params.precision)
+
+    def get_scheduler(self, scheduler_name:str, model: StableDiffusionGeneratorPipeline)->Scheduler:
+        scheduler_class = self.scheduler_map.get(scheduler_name,'ddim')
+        scheduler = scheduler_class.from_config(model.scheduler.config)
+        # hack copied over from generate.py
+        if not hasattr(scheduler, 'uses_inpainting_model'):
+            scheduler.uses_inpainting_model = lambda: False
+        return scheduler
+
+    @classmethod
+    def _generator_class(cls)->Type[Generator]:
+        '''
+        In derived classes return the name of the generator to apply.
+        If you don't override will return the name of the derived
+        class, which nicely parallels the generator class names.
+        '''
+        return Generator
+
+# ------------------------------------
+class Txt2Img(InvokeAIGenerator):
+    @classmethod
+    def _generator_class(cls):
+        from .txt2img import Txt2Img
+        return Txt2Img
+
+# ------------------------------------
+class Img2Img(InvokeAIGenerator):
+    def generate(self,
+               init_image: Image | torch.FloatTensor,
+               strength: float=0.75,
+               **keyword_args
+               )->List[InvokeAIGeneratorOutput]:
+        return super().generate(init_image=init_image,
+                                strength=strength,
+                                **keyword_args
+                                )
+    @classmethod
+    def _generator_class(cls):
+        from .img2img import Img2Img
+        return Img2Img
+
+# ------------------------------------
+# Takes all the arguments of Img2Img and adds the mask image and the seam/infill stuff
+class Inpaint(Img2Img):
+    def generate(self,
+                 mask_image: Image | torch.FloatTensor,
+                 # Seam settings - when 0, doesn't fill seam
+                 seam_size: int = 0,
+                 seam_blur: int = 0,
+                 seam_strength: float = 0.7,
+                 seam_steps: int = 10,
+                 tile_size: int = 32,
+                 inpaint_replace=False,
+                 infill_method=None,
+                 inpaint_width=None,
+                 inpaint_height=None,
+                 inpaint_fill: tuple(int) = (0x7F, 0x7F, 0x7F, 0xFF),
+                 **keyword_args
+                 )->List[InvokeAIGeneratorOutput]:
+        return super().generate(
+            mask_image=mask_image,
+            seam_size=seam_size,
+            seam_blur=seam_blur,
+            seam_strength=seam_strength,
+            seam_steps=seam_steps,
+            tile_size=tile_size,
+            inpaint_replace=inpaint_replace,
+            infill_method=infill_method,
+            inpaint_width=inpaint_width,
+            inpaint_height=inpaint_height,
+            inpaint_fill=inpaint_fill,
+            **keyword_args
+        )
+    @classmethod
+    def _generator_class(cls):
+        from .inpaint import Inpaint
+        return Inpaint
+
+# ------------------------------------
+class Embiggen(Txt2Img):
+    def generate(
+            self,
+            embiggen: list=None,
+            embiggen_tiles: list = None,
+            strength: float=0.75,
+            **kwargs)->List[InvokeAIGeneratorOutput]:
+        return super().generate(embiggen=embiggen,
+                                embiggen_tiles=embiggen_tiles,
+                                strength=strength,
+                                **kwargs)
+
+    @classmethod
+    def _generator_class(cls):
+        from .embiggen import Embiggen
+        return Embiggen
+
+
+class Generator:
+    downsampling_factor: int
+    latent_channels: int
+    precision: str
+    model: DiffusionPipeline
+
+    def __init__(self, model: DiffusionPipeline, precision: str):
+        self.model = model
+        self.precision = precision
+        self.seed = None
+        self.latent_channels = model.channels
+        self.downsampling_factor = downsampling  # BUG: should come from model or config
+        self.safety_checker = None
+        self.perlin = 0.0
+        self.threshold = 0
+        self.variation_amount = 0
+        self.with_variations = []
+        self.use_mps_noise = False
+        self.free_gpu_mem = None
+
+    # this is going to be overridden in img2img.py, txt2img.py and inpaint.py
+    def get_make_image(self, prompt, **kwargs):
+        """
+        Returns a function returning an image derived from the prompt and the initial image
+        Return value depends on the seed at the time you call it
+        """
+        raise NotImplementedError(
+            "image_iterator() must be implemented in a descendent class"
+        )
+
+    def set_variation(self, seed, variation_amount, with_variations):
+        self.seed = seed
+        self.variation_amount = variation_amount
+        self.with_variations = with_variations
+
+    def generate(
+        self,
+        prompt,
+        width,
+        height,
+        sampler,
+        init_image=None,
+        iterations=1,
+        seed=None,
+        image_callback=None,
+        step_callback=None,
+        threshold=0.0,
+        perlin=0.0,
+        h_symmetry_time_pct=None,
+        v_symmetry_time_pct=None,
+        safety_checker: SafetyChecker=None,
+        free_gpu_mem: bool = False,
+        **kwargs,
+    ):
+        scope = nullcontext
+        self.safety_checker = safety_checker
+        self.free_gpu_mem = free_gpu_mem
+        attention_maps_images = []
+        attention_maps_callback = lambda saver: attention_maps_images.append(
+            saver.get_stacked_maps_image()
+        )
+        make_image = self.get_make_image(
+            prompt,
+            sampler=sampler,
+            init_image=init_image,
+            width=width,
+            height=height,
+            step_callback=step_callback,
+            threshold=threshold,
+            perlin=perlin,
+            h_symmetry_time_pct=h_symmetry_time_pct,
+            v_symmetry_time_pct=v_symmetry_time_pct,
+            attention_maps_callback=attention_maps_callback,
+            **kwargs,
+        )
+        results = []
+        seed = seed if seed is not None and seed >= 0 else self.new_seed()
+        first_seed = seed
+        seed, initial_noise = self.generate_initial_noise(seed, width, height)
+
+        # There used to be an additional self.model.ema_scope() here, but it breaks
+        # the inpaint-1.5 model. Not sure what it did.... ?
+        with scope(self.model.device.type):
+            for n in trange(iterations, desc="Generating"):
+                x_T = None
+                if self.variation_amount > 0:
+                    set_seed(seed)
+                    target_noise = self.get_noise(width, height)
+                    x_T = self.slerp(self.variation_amount, initial_noise, target_noise)
+                elif initial_noise is not None:
+                    # i.e. we specified particular variations
+                    x_T = initial_noise
+                else:
+                    set_seed(seed)
+                    try:
+                        x_T = self.get_noise(width, height)
+                    except:
+                        print("** An error occurred while getting initial noise **")
+                        print(traceback.format_exc())
+
+                # Pass on the seed in case a layer beneath us needs to generate noise on its own.
+                image = make_image(x_T, seed)
+
+                if self.safety_checker is not None:
+                    image = self.safety_checker.check(image)
+
+                results.append([image, seed, attention_maps_images])
+
+                if image_callback is not None:
+                    attention_maps_image = (
+                        None
+                        if len(attention_maps_images) == 0
+                        else attention_maps_images[-1]
+                    )
+                    image_callback(
+                        image,
+                        seed,
+                        first_seed=first_seed,
+                        attention_maps_image=attention_maps_image,
+                    )
+
+                seed = self.new_seed()
+
+                # Free up memory from the last generation.
+                clear_cuda_cache = (
+                    kwargs["clear_cuda_cache"] if "clear_cuda_cache" in kwargs else None
+                )
+                if clear_cuda_cache is not None:
+                    clear_cuda_cache()
+
+        return results
+
+    def sample_to_image(self, samples) -> Image.Image:
+        """
+        Given samples returned from a sampler, converts
+        it into a PIL Image
+        """
+        with torch.inference_mode():
+            image = self.model.decode_latents(samples)
+        return self.model.numpy_to_pil(image)[0]
+
+    def repaste_and_color_correct(
+        self,
+        result: Image.Image,
+        init_image: Image.Image,
+        init_mask: Image.Image,
+        mask_blur_radius: int = 8,
+    ) -> Image.Image:
+        if init_image is None or init_mask is None:
+            return result
+
+        # Get the original alpha channel of the mask if there is one.
+        # Otherwise it is some other black/white image format ('1', 'L' or 'RGB')
+        pil_init_mask = (
+            init_mask.getchannel("A")
+            if init_mask.mode == "RGBA"
+            else init_mask.convert("L")
+        )
+        pil_init_image = init_image.convert(
+            "RGBA"
+        )  # Add an alpha channel if one doesn't exist
+
+        # Build an image with only visible pixels from source to use as reference for color-matching.
+        init_rgb_pixels = np.asarray(init_image.convert("RGB"), dtype=np.uint8)
+        init_a_pixels = np.asarray(pil_init_image.getchannel("A"), dtype=np.uint8)
+        init_mask_pixels = np.asarray(pil_init_mask, dtype=np.uint8)
+
+        # Get numpy version of result
+        np_image = np.asarray(result, dtype=np.uint8)
+
+        # Mask and calculate mean and standard deviation
+        mask_pixels = init_a_pixels * init_mask_pixels > 0
+        np_init_rgb_pixels_masked = init_rgb_pixels[mask_pixels, :]
+        np_image_masked = np_image[mask_pixels, :]
+
+        if np_init_rgb_pixels_masked.size > 0:
+            init_means = np_init_rgb_pixels_masked.mean(axis=0)
+            init_std = np_init_rgb_pixels_masked.std(axis=0)
+            gen_means = np_image_masked.mean(axis=0)
+            gen_std = np_image_masked.std(axis=0)
+
+            # Color correct
+            np_matched_result = np_image.copy()
+            np_matched_result[:, :, :] = (
+                (
+                    (
+                        (
+                            np_matched_result[:, :, :].astype(np.float32)
+                            - gen_means[None, None, :]
+                        )
+                        / gen_std[None, None, :]
+                    )
+                    * init_std[None, None, :]
+                    + init_means[None, None, :]
+                )
+                .clip(0, 255)
+                .astype(np.uint8)
+            )
+            matched_result = Image.fromarray(np_matched_result, mode="RGB")
+        else:
+            matched_result = Image.fromarray(np_image, mode="RGB")
+
+        # Blur the mask out (into init image) by specified amount
+        if mask_blur_radius > 0:
+            nm = np.asarray(pil_init_mask, dtype=np.uint8)
+            nmd = cv2.erode(
+                nm,
+                kernel=np.ones((3, 3), dtype=np.uint8),
+                iterations=int(mask_blur_radius / 2),
+            )
+            pmd = Image.fromarray(nmd, mode="L")
+            blurred_init_mask = pmd.filter(ImageFilter.BoxBlur(mask_blur_radius))
+        else:
+            blurred_init_mask = pil_init_mask
+
+        multiplied_blurred_init_mask = ImageChops.multiply(
+            blurred_init_mask, self.pil_image.split()[-1]
+        )
+
+        # Paste original on color-corrected generation (using blurred mask)
+        matched_result.paste(init_image, (0, 0), mask=multiplied_blurred_init_mask)
+        return matched_result
+
+    @staticmethod
+    def sample_to_lowres_estimated_image(samples):
+        # origingally adapted from code by @erucipe and @keturn here:
+        # https://discuss.huggingface.co/t/decoding-latents-to-rgb-without-upscaling/23204/7
+
+        # these updated numbers for v1.5 are from @torridgristle
+        v1_5_latent_rgb_factors = torch.tensor(
+            [
+                #    R        G        B
+                [0.3444, 0.1385, 0.0670],  # L1
+                [0.1247, 0.4027, 0.1494],  # L2
+                [-0.3192, 0.2513, 0.2103],  # L3
+                [-0.1307, -0.1874, -0.7445],  # L4
+            ],
+            dtype=samples.dtype,
+            device=samples.device,
+        )
+
+        latent_image = samples[0].permute(1, 2, 0) @ v1_5_latent_rgb_factors
+        latents_ubyte = (
+            ((latent_image + 1) / 2)
+            .clamp(0, 1)  # change scale from -1..1 to 0..1
+            .mul(0xFF)  # to 0..255
+            .byte()
+        ).cpu()
+
+        return Image.fromarray(latents_ubyte.numpy())
+
+    def generate_initial_noise(self, seed, width, height):
+        initial_noise = None
+        if self.variation_amount > 0 or len(self.with_variations) > 0:
+            # use fixed initial noise plus random noise per iteration
+            set_seed(seed)
+            initial_noise = self.get_noise(width, height)
+            for v_seed, v_weight in self.with_variations:
+                seed = v_seed
+                set_seed(seed)
+                next_noise = self.get_noise(width, height)
+                initial_noise = self.slerp(v_weight, initial_noise, next_noise)
+            if self.variation_amount > 0:
+                random.seed()  # reset RNG to an actually random state, so we can get a random seed for variations
+                seed = random.randrange(0, np.iinfo(np.uint32).max)
+        return (seed, initial_noise)
+
+    def get_perlin_noise(self, width, height):
+        fixdevice = "cpu" if (self.model.device.type == "mps") else self.model.device
+        # limit noise to only the diffusion image channels, not the mask channels
+        input_channels = min(self.latent_channels, 4)
+        # round up to the nearest block of 8
+        temp_width = int((width + 7) / 8) * 8
+        temp_height = int((height + 7) / 8) * 8
+        noise = torch.stack(
+            [
+                rand_perlin_2d(
+                    (temp_height, temp_width), (8, 8), device=self.model.device
+                ).to(fixdevice)
+                for _ in range(input_channels)
+            ],
+            dim=0,
+        ).to(self.model.device)
+        return noise[0:4, 0:height, 0:width]
+
+    def new_seed(self):
+        self.seed = random.randrange(0, np.iinfo(np.uint32).max)
+        return self.seed
+
+    def slerp(self, t, v0, v1, DOT_THRESHOLD=0.9995):
+        """
+        Spherical linear interpolation
+        Args:
+            t (float/np.ndarray): Float value between 0.0 and 1.0
+            v0 (np.ndarray): Starting vector
+            v1 (np.ndarray): Final vector
+            DOT_THRESHOLD (float): Threshold for considering the two vectors as
+                                colineal. Not recommended to alter this.
+        Returns:
+            v2 (np.ndarray): Interpolation vector between v0 and v1
+        """
+        inputs_are_torch = False
+        if not isinstance(v0, np.ndarray):
+            inputs_are_torch = True
+            v0 = v0.detach().cpu().numpy()
+        if not isinstance(v1, np.ndarray):
+            inputs_are_torch = True
+            v1 = v1.detach().cpu().numpy()
+
+        dot = np.sum(v0 * v1 / (np.linalg.norm(v0) * np.linalg.norm(v1)))
+        if np.abs(dot) > DOT_THRESHOLD:
+            v2 = (1 - t) * v0 + t * v1
+        else:
+            theta_0 = np.arccos(dot)
+            sin_theta_0 = np.sin(theta_0)
+            theta_t = theta_0 * t
+            sin_theta_t = np.sin(theta_t)
+            s0 = np.sin(theta_0 - theta_t) / sin_theta_0
+            s1 = sin_theta_t / sin_theta_0
+            v2 = s0 * v0 + s1 * v1
+
+        if inputs_are_torch:
+            v2 = torch.from_numpy(v2).to(self.model.device)
+
+        return v2
+
+    # this is a handy routine for debugging use. Given a generated sample,
+    # convert it into a PNG image and store it at the indicated path
+    def save_sample(self, sample, filepath):
+        image = self.sample_to_image(sample)
+        dirname = os.path.dirname(filepath) or "."
+        if not os.path.exists(dirname):
+            print(f"** creating directory {dirname}")
+            os.makedirs(dirname, exist_ok=True)
+        image.save(filepath, "PNG")
+
+    def torch_dtype(self) -> torch.dtype:
+        return torch.float16 if self.precision == "float16" else torch.float32
+
+    # returns a tensor filled with random numbers from a normal distribution
+    def get_noise(self, width, height):
+        device = self.model.device
+        # limit noise to only the diffusion image channels, not the mask channels
+        input_channels = min(self.latent_channels, 4)
+        if self.use_mps_noise or device.type == "mps":
+            x = torch.randn(
+                [
+                    1,
+                    input_channels,
+                    height // self.downsampling_factor,
+                    width // self.downsampling_factor,
+                ],
+                dtype=self.torch_dtype(),
+                device="cpu",
+            ).to(device)
+        else:
+            x = torch.randn(
+                [
+                    1,
+                    input_channels,
+                    height // self.downsampling_factor,
+                    width // self.downsampling_factor,
+                ],
+                dtype=self.torch_dtype(),
+                device=device,
+            )
+        if self.perlin > 0.0:
+            perlin_noise = self.get_perlin_noise(
+                width // self.downsampling_factor, height // self.downsampling_factor
+            )
+            x = (1 - self.perlin) * x + self.perlin * perlin_noise
+        return x

invokeai/backend/generator/embiggen.py

@@ -1,37 +1,38 @@
-'''
-ldm.invoke.generator.embiggen descends from ldm.invoke.generator
-and generates with ldm.invoke.generator.img2img
-'''
+"""
+invokeai.backend.generator.embiggen descends from .generator
+and generates with .generator.img2img
+"""
 
 import numpy as np
 import torch
 from PIL import Image
 from tqdm import trange
 
-from ldm.invoke.generator.base import Generator
-from ldm.invoke.generator.img2img import Img2Img
+from .base import Generator
+from .img2img import Img2Img
 
 
 class Embiggen(Generator):
     def __init__(self, model, precision):
         super().__init__(model, precision)
-        self.init_latent         = None
+        self.init_latent = None
 
     # Replace generate because Embiggen doesn't need/use most of what it does normallly
-    def generate(self,prompt,iterations=1,seed=None,
-                 image_callback=None, step_callback=None,
-                 **kwargs):
-
-        make_image = self.get_make_image(
-            prompt,
-            step_callback = step_callback,
-            **kwargs
-        )
-        results             = []
-        seed                = seed if seed else self.new_seed()
+    def generate(
+        self,
+        prompt,
+        iterations=1,
+        seed=None,
+        image_callback=None,
+        step_callback=None,
+        **kwargs,
+    ):
+        make_image = self.get_make_image(prompt, step_callback=step_callback, **kwargs)
+        results = []
+        seed = seed if seed else self.new_seed()
 
         # Noise will be generated by the Img2Img generator when called
-        for _ in trange(iterations, desc='Generating'):
+        for _ in trange(iterations, desc="Generating"):
             # make_image will call Img2Img which will do the equivalent of get_noise itself
             image = make_image()
             results.append([image, seed])
@@ -56,13 +57,15 @@ class Embiggen(Generator):
         embiggen,
         embiggen_tiles,
         step_callback=None,
-        **kwargs
+        **kwargs,
     ):
         """
         Returns a function returning an image derived from the prompt and multi-stage twice-baked potato layering over the img2img on the initial image
         Return value depends on the seed at the time you call it
         """
-        assert not sampler.uses_inpainting_model(), "--embiggen is not supported by inpainting models"
+        assert (
+            not sampler.uses_inpainting_model()
+        ), "--embiggen is not supported by inpainting models"
 
         # Construct embiggen arg array, and sanity check arguments
         if embiggen == None:  # embiggen can also be called with just embiggen_tiles
@@ -70,48 +73,57 @@ class Embiggen(Generator):
         elif embiggen[0] < 0:
             embiggen[0] = 1.0
             print(
-                '>> Embiggen scaling factor cannot be negative, fell back to the default of 1.0 !')
+                ">> Embiggen scaling factor cannot be negative, fell back to the default of 1.0 !"
+            )
         if len(embiggen) < 2:
             embiggen.append(0.75)
         elif embiggen[1] > 1.0 or embiggen[1] < 0:
             embiggen[1] = 0.75
-            print('>> Embiggen upscaling strength for ESRGAN must be between 0 and 1, fell back to the default of 0.75 !')
+            print(
+                ">> Embiggen upscaling strength for ESRGAN must be between 0 and 1, fell back to the default of 0.75 !"
+            )
         if len(embiggen) < 3:
             embiggen.append(0.25)
         elif embiggen[2] < 0:
             embiggen[2] = 0.25
-            print('>> Overlap size for Embiggen must be a positive ratio between 0 and 1 OR a number of pixels, fell back to the default of 0.25 !')
+            print(
+                ">> Overlap size for Embiggen must be a positive ratio between 0 and 1 OR a number of pixels, fell back to the default of 0.25 !"
+            )
 
         # Convert tiles from their user-freindly count-from-one to count-from-zero, because we need to do modulo math
         # and then sort them, because... people.
         if embiggen_tiles:
-            embiggen_tiles = list(map(lambda n: n-1, embiggen_tiles))
+            embiggen_tiles = list(map(lambda n: n - 1, embiggen_tiles))
             embiggen_tiles.sort()
 
         if strength >= 0.5:
-            print(f'* WARNING: Embiggen may produce mirror motifs if the strength (-f) is too high (currently {strength}). Try values between 0.35-0.45.')
+            print(
+                f"* WARNING: Embiggen may produce mirror motifs if the strength (-f) is too high (currently {strength}). Try values between 0.35-0.45."
+            )
 
         # Prep img2img generator, since we wrap over it
-        gen_img2img = Img2Img(self.model,self.precision)
+        gen_img2img = Img2Img(self.model, self.precision)
 
         # Open original init image (not a tensor) to manipulate
         initsuperimage = Image.open(init_img)
 
         with Image.open(init_img) as img:
-            initsuperimage = img.convert('RGB')
+            initsuperimage = img.convert("RGB")
 
         # Size of the target super init image in pixels
         initsuperwidth, initsuperheight = initsuperimage.size
 
         # Increase by scaling factor if not already resized, using ESRGAN as able
         if embiggen[0] != 1.0:
-            initsuperwidth = round(initsuperwidth*embiggen[0])
-            initsuperheight = round(initsuperheight*embiggen[0])
+            initsuperwidth = round(initsuperwidth * embiggen[0])
+            initsuperheight = round(initsuperheight * embiggen[0])
             if embiggen[1] > 0:  # No point in ESRGAN upscaling if strength is set zero
-                from ldm.invoke.restoration.realesrgan import ESRGAN
+                from ..restoration.realesrgan import ESRGAN
+
                 esrgan = ESRGAN()
                 print(
-                    f'>> ESRGAN upscaling init image prior to cutting with Embiggen with strength {embiggen[1]}')
+                    f">> ESRGAN upscaling init image prior to cutting with Embiggen with strength {embiggen[1]}"
+                )
                 if embiggen[0] > 2:
                     initsuperimage = esrgan.process(
                         initsuperimage,
@@ -130,7 +142,8 @@ class Embiggen(Generator):
             #   but from personal experiance it doesn't greatly improve anything after 4x
             # Resize to target scaling factor resolution
             initsuperimage = initsuperimage.resize(
-                (initsuperwidth, initsuperheight), Image.Resampling.LANCZOS)
+                (initsuperwidth, initsuperheight), Image.Resampling.LANCZOS
+            )
 
         # Use width and height as tile widths and height
         # Determine buffer size in pixels
@@ -153,23 +166,24 @@ class Embiggen(Generator):
         emb_tiles_x = 1
         emb_tiles_y = 1
         if (initsuperwidth - width) > 0:
-            emb_tiles_x = ceildiv(initsuperwidth - width,
-                                  width - overlap_size_x) + 1
+            emb_tiles_x = ceildiv(initsuperwidth - width, width - overlap_size_x) + 1
         if (initsuperheight - height) > 0:
-            emb_tiles_y = ceildiv(initsuperheight - height,
-                                  height - overlap_size_y) + 1
+            emb_tiles_y = ceildiv(initsuperheight - height, height - overlap_size_y) + 1
         # Sanity
-        assert emb_tiles_x > 1 or emb_tiles_y > 1, f'ERROR: Based on the requested dimensions of {initsuperwidth}x{initsuperheight} and tiles of {width}x{height} you don\'t need to Embiggen! Check your arguments.'
+        assert (
+            emb_tiles_x > 1 or emb_tiles_y > 1
+        ), f"ERROR: Based on the requested dimensions of {initsuperwidth}x{initsuperheight} and tiles of {width}x{height} you don't need to Embiggen! Check your arguments."
 
         # Prep alpha layers --------------
         # https://stackoverflow.com/questions/69321734/how-to-create-different-transparency-like-gradient-with-python-pil
         # agradientL is Left-side transparent
-        agradientL = Image.linear_gradient('L').rotate(
-            90).resize((overlap_size_x, height))
+        agradientL = (
+            Image.linear_gradient("L").rotate(90).resize((overlap_size_x, height))
+        )
         # agradientT is Top-side transparent
-        agradientT = Image.linear_gradient('L').resize((width, overlap_size_y))
+        agradientT = Image.linear_gradient("L").resize((width, overlap_size_y))
         # radial corner is the left-top corner, made full circle then cut to just the left-top quadrant
-        agradientC = Image.new('L', (256, 256))
+        agradientC = Image.new("L", (256, 256))
         for y in range(256):
             for x in range(256):
                 # Find distance to lower right corner (numpy takes arrays)
@@ -177,16 +191,16 @@ class Embiggen(Generator):
                 # Clamp values to max 255
                 if distanceToLR > 255:
                     distanceToLR = 255
-                #Place the pixel as invert of distance
+                # Place the pixel as invert of distance
                 agradientC.putpixel((x, y), round(255 - distanceToLR))
 
         # Create alternative asymmetric diagonal corner to use on "tailing" intersections to prevent hard edges
         # Fits for a left-fading gradient on the bottom side and full opacity on the right side.
-        agradientAsymC = Image.new('L', (256, 256))
+        agradientAsymC = Image.new("L", (256, 256))
         for y in range(256):
             for x in range(256):
-                value = round(max(0, x-(255-y)) * (255 / max(1,y)))
-                #Clamp values
+                value = round(max(0, x - (255 - y)) * (255 / max(1, y)))
+                # Clamp values
                 value = max(0, value)
                 value = min(255, value)
                 agradientAsymC.putpixel((x, y), value)
@@ -204,80 +218,91 @@ class Embiggen(Generator):
         # make masks with an asymmetric upper-right corner so when the curved transparent corner of the next tile
         # to its right is placed it doesn't reveal a hard trailing semi-transparent edge in the overlapping space
         alphaLayerTaC = alphaLayerT.copy()
-        alphaLayerTaC.paste(agradientAsymC.rotate(270).resize((overlap_size_x, overlap_size_y)), (width - overlap_size_x, 0))
+        alphaLayerTaC.paste(
+            agradientAsymC.rotate(270).resize((overlap_size_x, overlap_size_y)),
+            (width - overlap_size_x, 0),
+        )
         alphaLayerLTaC = alphaLayerLTC.copy()
-        alphaLayerLTaC.paste(agradientAsymC.rotate(270).resize((overlap_size_x, overlap_size_y)), (width - overlap_size_x, 0))
+        alphaLayerLTaC.paste(
+            agradientAsymC.rotate(270).resize((overlap_size_x, overlap_size_y)),
+            (width - overlap_size_x, 0),
+        )
 
         if embiggen_tiles:
             # Individual unconnected sides
             alphaLayerR = Image.new("L", (width, height), 255)
-            alphaLayerR.paste(agradientL.rotate(
-                180), (width - overlap_size_x, 0))
+            alphaLayerR.paste(agradientL.rotate(180), (width - overlap_size_x, 0))
             alphaLayerB = Image.new("L", (width, height), 255)
-            alphaLayerB.paste(agradientT.rotate(
-                180), (0, height - overlap_size_y))
+            alphaLayerB.paste(agradientT.rotate(180), (0, height - overlap_size_y))
             alphaLayerTB = Image.new("L", (width, height), 255)
             alphaLayerTB.paste(agradientT, (0, 0))
-            alphaLayerTB.paste(agradientT.rotate(
-                180), (0, height - overlap_size_y))
+            alphaLayerTB.paste(agradientT.rotate(180), (0, height - overlap_size_y))
             alphaLayerLR = Image.new("L", (width, height), 255)
             alphaLayerLR.paste(agradientL, (0, 0))
-            alphaLayerLR.paste(agradientL.rotate(
-                180), (width - overlap_size_x, 0))
+            alphaLayerLR.paste(agradientL.rotate(180), (width - overlap_size_x, 0))
 
             # Sides and corner Layers
             alphaLayerRBC = Image.new("L", (width, height), 255)
-            alphaLayerRBC.paste(agradientL.rotate(
-                180), (width - overlap_size_x, 0))
-            alphaLayerRBC.paste(agradientT.rotate(
-                180), (0, height - overlap_size_y))
-            alphaLayerRBC.paste(agradientC.rotate(180).resize(
-                (overlap_size_x, overlap_size_y)), (width - overlap_size_x, height - overlap_size_y))
+            alphaLayerRBC.paste(agradientL.rotate(180), (width - overlap_size_x, 0))
+            alphaLayerRBC.paste(agradientT.rotate(180), (0, height - overlap_size_y))
+            alphaLayerRBC.paste(
+                agradientC.rotate(180).resize((overlap_size_x, overlap_size_y)),
+                (width - overlap_size_x, height - overlap_size_y),
+            )
             alphaLayerLBC = Image.new("L", (width, height), 255)
             alphaLayerLBC.paste(agradientL, (0, 0))
-            alphaLayerLBC.paste(agradientT.rotate(
-                180), (0, height - overlap_size_y))
-            alphaLayerLBC.paste(agradientC.rotate(90).resize(
-                (overlap_size_x, overlap_size_y)), (0, height - overlap_size_y))
+            alphaLayerLBC.paste(agradientT.rotate(180), (0, height - overlap_size_y))
+            alphaLayerLBC.paste(
+                agradientC.rotate(90).resize((overlap_size_x, overlap_size_y)),
+                (0, height - overlap_size_y),
+            )
             alphaLayerRTC = Image.new("L", (width, height), 255)
-            alphaLayerRTC.paste(agradientL.rotate(
-                180), (width - overlap_size_x, 0))
+            alphaLayerRTC.paste(agradientL.rotate(180), (width - overlap_size_x, 0))
             alphaLayerRTC.paste(agradientT, (0, 0))
-            alphaLayerRTC.paste(agradientC.rotate(270).resize(
-                (overlap_size_x, overlap_size_y)), (width - overlap_size_x, 0))
+            alphaLayerRTC.paste(
+                agradientC.rotate(270).resize((overlap_size_x, overlap_size_y)),
+                (width - overlap_size_x, 0),
+            )
 
             # All but X layers
             alphaLayerABT = Image.new("L", (width, height), 255)
             alphaLayerABT.paste(alphaLayerLBC, (0, 0))
-            alphaLayerABT.paste(agradientL.rotate(
-                180), (width - overlap_size_x, 0))
-            alphaLayerABT.paste(agradientC.rotate(180).resize(
-                (overlap_size_x, overlap_size_y)), (width - overlap_size_x, height - overlap_size_y))
+            alphaLayerABT.paste(agradientL.rotate(180), (width - overlap_size_x, 0))
+            alphaLayerABT.paste(
+                agradientC.rotate(180).resize((overlap_size_x, overlap_size_y)),
+                (width - overlap_size_x, height - overlap_size_y),
+            )
             alphaLayerABL = Image.new("L", (width, height), 255)
             alphaLayerABL.paste(alphaLayerRTC, (0, 0))
-            alphaLayerABL.paste(agradientT.rotate(
-                180), (0, height - overlap_size_y))
-            alphaLayerABL.paste(agradientC.rotate(180).resize(
-                (overlap_size_x, overlap_size_y)), (width - overlap_size_x, height - overlap_size_y))
+            alphaLayerABL.paste(agradientT.rotate(180), (0, height - overlap_size_y))
+            alphaLayerABL.paste(
+                agradientC.rotate(180).resize((overlap_size_x, overlap_size_y)),
+                (width - overlap_size_x, height - overlap_size_y),
+            )
             alphaLayerABR = Image.new("L", (width, height), 255)
             alphaLayerABR.paste(alphaLayerLBC, (0, 0))
             alphaLayerABR.paste(agradientT, (0, 0))
-            alphaLayerABR.paste(agradientC.resize(
-                (overlap_size_x, overlap_size_y)), (0, 0))
+            alphaLayerABR.paste(
+                agradientC.resize((overlap_size_x, overlap_size_y)), (0, 0)
+            )
             alphaLayerABB = Image.new("L", (width, height), 255)
             alphaLayerABB.paste(alphaLayerRTC, (0, 0))
             alphaLayerABB.paste(agradientL, (0, 0))
-            alphaLayerABB.paste(agradientC.resize(
-                (overlap_size_x, overlap_size_y)), (0, 0))
+            alphaLayerABB.paste(
+                agradientC.resize((overlap_size_x, overlap_size_y)), (0, 0)
+            )
 
             # All-around layer
             alphaLayerAA = Image.new("L", (width, height), 255)
             alphaLayerAA.paste(alphaLayerABT, (0, 0))
             alphaLayerAA.paste(agradientT, (0, 0))
-            alphaLayerAA.paste(agradientC.resize(
-                (overlap_size_x, overlap_size_y)), (0, 0))
-            alphaLayerAA.paste(agradientC.rotate(270).resize(
-                (overlap_size_x, overlap_size_y)), (width - overlap_size_x, 0))
+            alphaLayerAA.paste(
+                agradientC.resize((overlap_size_x, overlap_size_y)), (0, 0)
+            )
+            alphaLayerAA.paste(
+                agradientC.rotate(270).resize((overlap_size_x, overlap_size_y)),
+                (width - overlap_size_x, 0),
+            )
 
         # Clean up temporary gradients
         del agradientL
@@ -287,17 +312,20 @@ class Embiggen(Generator):
         def make_image():
             # Make main tiles -------------------------------------------------
             if embiggen_tiles:
-                print(f'>> Making {len(embiggen_tiles)} Embiggen tiles...')
+                print(f">> Making {len(embiggen_tiles)} Embiggen tiles...")
             else:
                 print(
-                    f'>> Making {(emb_tiles_x * emb_tiles_y)} Embiggen tiles ({emb_tiles_x}x{emb_tiles_y})...')
+                    f">> Making {(emb_tiles_x * emb_tiles_y)} Embiggen tiles ({emb_tiles_x}x{emb_tiles_y})..."
+                )
 
             emb_tile_store = []
             # Although we could use the same seed for every tile for determinism, at higher strengths this may
             # produce duplicated structures for each tile and make the tiling effect more obvious
             # instead track and iterate a local seed we pass to Img2Img
             seed = self.seed
-            seedintlimit = np.iinfo(np.uint32).max - 1 # only retreive this one from numpy
+            seedintlimit = (
+                np.iinfo(np.uint32).max - 1
+            )  # only retreive this one from numpy
 
             for tile in range(emb_tiles_x * emb_tiles_y):
                 # Don't iterate on first tile
@@ -334,37 +362,38 @@ class Embiggen(Generator):
 
                 if embiggen_tiles:
                     print(
-                        f'Making tile #{tile + 1} ({embiggen_tiles.index(tile) + 1} of {len(embiggen_tiles)} requested)')
+                        f"Making tile #{tile + 1} ({embiggen_tiles.index(tile) + 1} of {len(embiggen_tiles)} requested)"
+                    )
                 else:
-                    print(
-                        f'Starting {tile + 1} of {(emb_tiles_x * emb_tiles_y)} tiles')
+                    print(f"Starting {tile + 1} of {(emb_tiles_x * emb_tiles_y)} tiles")
 
                 # create a torch tensor from an Image
-                newinitimage = np.array(
-                    newinitimage).astype(np.float32) / 255.0
+                newinitimage = np.array(newinitimage).astype(np.float32) / 255.0
                 newinitimage = newinitimage[None].transpose(0, 3, 1, 2)
                 newinitimage = torch.from_numpy(newinitimage)
                 newinitimage = 2.0 * newinitimage - 1.0
                 newinitimage = newinitimage.to(self.model.device)
-                clear_cuda_cache = kwargs['clear_cuda_cache'] if 'clear_cuda_cache' in kwargs else None
+                clear_cuda_cache = (
+                    kwargs["clear_cuda_cache"] if "clear_cuda_cache" in kwargs else None
+                )
 
                 tile_results = gen_img2img.generate(
                     prompt,
-                    iterations     = 1,
-                    seed           = seed,
-                    sampler        = sampler,
-                    steps          = steps,
-                    cfg_scale      = cfg_scale,
-                    conditioning   = conditioning,
-                    ddim_eta       = ddim_eta,
-                    image_callback = None,  # called only after the final image is generated
-                    step_callback  = step_callback,   # called after each intermediate image is generated
-                    width          = width,
-                    height         = height,
-                    init_image     = newinitimage,    # notice that init_image is different from init_img
-                    mask_image     = None,
-                    strength       = strength,
-                    clear_cuda_cache     = clear_cuda_cache
+                    iterations=1,
+                    seed=seed,
+                    sampler=sampler,
+                    steps=steps,
+                    cfg_scale=cfg_scale,
+                    conditioning=conditioning,
+                    ddim_eta=ddim_eta,
+                    image_callback=None,  # called only after the final image is generated
+                    step_callback=step_callback,  # called after each intermediate image is generated
+                    width=width,
+                    height=height,
+                    init_image=newinitimage,  # notice that init_image is different from init_img
+                    mask_image=None,
+                    strength=strength,
+                    clear_cuda_cache=clear_cuda_cache,
                 )
 
                 emb_tile_store.append(tile_results[0][0])
@@ -373,12 +402,14 @@ class Embiggen(Generator):
                 del newinitimage
 
             # Sanity check we have them all
-            if len(emb_tile_store) == (emb_tiles_x * emb_tiles_y) or (embiggen_tiles != [] and len(emb_tile_store) == len(embiggen_tiles)):
-                outputsuperimage = Image.new(
-                    "RGBA", (initsuperwidth, initsuperheight))
+            if len(emb_tile_store) == (emb_tiles_x * emb_tiles_y) or (
+                embiggen_tiles != [] and len(emb_tile_store) == len(embiggen_tiles)
+            ):
+                outputsuperimage = Image.new("RGBA", (initsuperwidth, initsuperheight))
                 if embiggen_tiles:
                     outputsuperimage.alpha_composite(
-                        initsuperimage.convert('RGBA'), (0, 0))
+                        initsuperimage.convert("RGBA"), (0, 0)
+                    )
                 for tile in range(emb_tiles_x * emb_tiles_y):
                     if embiggen_tiles:
                         if tile in embiggen_tiles:
@@ -387,7 +418,7 @@ class Embiggen(Generator):
                             continue
                     else:
                         intileimage = emb_tile_store[tile]
-                    intileimage = intileimage.convert('RGBA')
+                    intileimage = intileimage.convert("RGBA")
                     # Get row and column entries
                     emb_row_i = tile // emb_tiles_x
                     emb_column_i = tile % emb_tiles_x
@@ -399,8 +430,7 @@ class Embiggen(Generator):
                         if emb_column_i + 1 == emb_tiles_x:
                             left = initsuperwidth - width
                         else:
-                            left = round(emb_column_i *
-                                         (width - overlap_size_x))
+                            left = round(emb_column_i * (width - overlap_size_x))
                         if emb_row_i + 1 == emb_tiles_y:
                             top = initsuperheight - height
                         else:
@@ -411,33 +441,43 @@ class Embiggen(Generator):
                             # top of image
                             if emb_row_i == 0:
                                 if emb_column_i == 0:
-                                    if (tile+1) in embiggen_tiles:  # Look-ahead right
-                                        if (tile+emb_tiles_x) not in embiggen_tiles:  # Look-ahead down
+                                    if (tile + 1) in embiggen_tiles:  # Look-ahead right
+                                        if (
+                                            tile + emb_tiles_x
+                                        ) not in embiggen_tiles:  # Look-ahead down
                                             intileimage.putalpha(alphaLayerB)
                                         # Otherwise do nothing on this tile
-                                    elif (tile+emb_tiles_x) in embiggen_tiles:  # Look-ahead down only
+                                    elif (
+                                        tile + emb_tiles_x
+                                    ) in embiggen_tiles:  # Look-ahead down only
                                         intileimage.putalpha(alphaLayerR)
                                     else:
                                         intileimage.putalpha(alphaLayerRBC)
                                 elif emb_column_i == emb_tiles_x - 1:
-                                    if (tile+emb_tiles_x) in embiggen_tiles:  # Look-ahead down
+                                    if (
+                                        tile + emb_tiles_x
+                                    ) in embiggen_tiles:  # Look-ahead down
                                         intileimage.putalpha(alphaLayerL)
                                     else:
                                         intileimage.putalpha(alphaLayerLBC)
                                 else:
-                                    if (tile+1) in embiggen_tiles:  # Look-ahead right
-                                        if (tile+emb_tiles_x) in embiggen_tiles:  # Look-ahead down
+                                    if (tile + 1) in embiggen_tiles:  # Look-ahead right
+                                        if (
+                                            tile + emb_tiles_x
+                                        ) in embiggen_tiles:  # Look-ahead down
                                             intileimage.putalpha(alphaLayerL)
                                         else:
                                             intileimage.putalpha(alphaLayerLBC)
-                                    elif (tile+emb_tiles_x) in embiggen_tiles:  # Look-ahead down only
+                                    elif (
+                                        tile + emb_tiles_x
+                                    ) in embiggen_tiles:  # Look-ahead down only
                                         intileimage.putalpha(alphaLayerLR)
                                     else:
                                         intileimage.putalpha(alphaLayerABT)
                             # bottom of image
                             elif emb_row_i == emb_tiles_y - 1:
                                 if emb_column_i == 0:
-                                    if (tile+1) in embiggen_tiles: # Look-ahead right
+                                    if (tile + 1) in embiggen_tiles:  # Look-ahead right
                                         intileimage.putalpha(alphaLayerTaC)
                                     else:
                                         intileimage.putalpha(alphaLayerRTC)
@@ -445,34 +485,44 @@ class Embiggen(Generator):
                                     # No tiles to look ahead to
                                     intileimage.putalpha(alphaLayerLTC)
                                 else:
-                                    if (tile+1) in embiggen_tiles: # Look-ahead right
+                                    if (tile + 1) in embiggen_tiles:  # Look-ahead right
                                         intileimage.putalpha(alphaLayerLTaC)
                                     else:
                                         intileimage.putalpha(alphaLayerABB)
                             # vertical middle of image
                             else:
                                 if emb_column_i == 0:
-                                    if (tile+1) in embiggen_tiles: # Look-ahead right
-                                        if (tile+emb_tiles_x) in embiggen_tiles: # Look-ahead down
+                                    if (tile + 1) in embiggen_tiles:  # Look-ahead right
+                                        if (
+                                            tile + emb_tiles_x
+                                        ) in embiggen_tiles:  # Look-ahead down
                                             intileimage.putalpha(alphaLayerTaC)
                                         else:
                                             intileimage.putalpha(alphaLayerTB)
-                                    elif (tile+emb_tiles_x) in embiggen_tiles:  # Look-ahead down only
+                                    elif (
+                                        tile + emb_tiles_x
+                                    ) in embiggen_tiles:  # Look-ahead down only
                                         intileimage.putalpha(alphaLayerRTC)
                                     else:
                                         intileimage.putalpha(alphaLayerABL)
                                 elif emb_column_i == emb_tiles_x - 1:
-                                    if (tile+emb_tiles_x) in embiggen_tiles:  # Look-ahead down
+                                    if (
+                                        tile + emb_tiles_x
+                                    ) in embiggen_tiles:  # Look-ahead down
                                         intileimage.putalpha(alphaLayerLTC)
                                     else:
                                         intileimage.putalpha(alphaLayerABR)
                                 else:
-                                    if (tile+1) in embiggen_tiles: # Look-ahead right
-                                        if (tile+emb_tiles_x) in embiggen_tiles: # Look-ahead down
+                                    if (tile + 1) in embiggen_tiles:  # Look-ahead right
+                                        if (
+                                            tile + emb_tiles_x
+                                        ) in embiggen_tiles:  # Look-ahead down
                                             intileimage.putalpha(alphaLayerLTaC)
                                         else:
                                             intileimage.putalpha(alphaLayerABR)
-                                    elif (tile+emb_tiles_x) in embiggen_tiles:  # Look-ahead down only
+                                    elif (
+                                        tile + emb_tiles_x
+                                    ) in embiggen_tiles:  # Look-ahead down only
                                         intileimage.putalpha(alphaLayerABB)
                                     else:
                                         intileimage.putalpha(alphaLayerAA)
@@ -481,21 +531,28 @@ class Embiggen(Generator):
                             if emb_row_i == 0 and emb_column_i >= 1:
                                 intileimage.putalpha(alphaLayerL)
                             elif emb_row_i >= 1 and emb_column_i == 0:
-                                if emb_column_i + 1 == emb_tiles_x: # If we don't have anything that can be placed to the right
+                                if (
+                                    emb_column_i + 1 == emb_tiles_x
+                                ):  # If we don't have anything that can be placed to the right
                                     intileimage.putalpha(alphaLayerT)
                                 else:
                                     intileimage.putalpha(alphaLayerTaC)
                             else:
-                                if emb_column_i + 1 == emb_tiles_x: # If we don't have anything that can be placed to the right
+                                if (
+                                    emb_column_i + 1 == emb_tiles_x
+                                ):  # If we don't have anything that can be placed to the right
                                     intileimage.putalpha(alphaLayerLTC)
                                 else:
                                     intileimage.putalpha(alphaLayerLTaC)
                     # Layer tile onto final image
                     outputsuperimage.alpha_composite(intileimage, (left, top))
             else:
-                print('Error: could not find all Embiggen output tiles in memory? Something must have gone wrong with img2img generation.')
+                print(
+                    "Error: could not find all Embiggen output tiles in memory? Something must have gone wrong with img2img generation."
+                )
 
             # after internal loops and patching up return Embiggen image
             return outputsuperimage
+
         # end of function declaration
         return make_image

invokeai/backend/generator/img2img.py

@@ -0,0 +1,101 @@
+"""
+invokeai.backend.generator.img2img descends from .generator
+"""
+from typing import Optional
+
+import torch
+from accelerate.utils import set_seed
+from diffusers import logging
+
+from ..stable_diffusion import (
+    ConditioningData,
+    PostprocessingSettings,
+    StableDiffusionGeneratorPipeline,
+)
+from .base import Generator
+
+
+class Img2Img(Generator):
+    def __init__(self, model, precision):
+        super().__init__(model, precision)
+        self.init_latent = None  # by get_noise()
+
+    def get_make_image(
+        self,
+        prompt,
+        sampler,
+        steps,
+        cfg_scale,
+        ddim_eta,
+        conditioning,
+        init_image,
+        strength,
+        step_callback=None,
+        threshold=0.0,
+        warmup=0.2,
+        perlin=0.0,
+        h_symmetry_time_pct=None,
+        v_symmetry_time_pct=None,
+        attention_maps_callback=None,
+        **kwargs,
+    ):
+        """
+        Returns a function returning an image derived from the prompt and the initial image
+        Return value depends on the seed at the time you call it.
+        """
+        self.perlin = perlin
+
+        # noinspection PyTypeChecker
+        pipeline: StableDiffusionGeneratorPipeline = self.model
+        pipeline.scheduler = sampler
+
+        uc, c, extra_conditioning_info = conditioning
+        conditioning_data = ConditioningData(
+            uc,
+            c,
+            cfg_scale,
+            extra_conditioning_info,
+            postprocessing_settings=PostprocessingSettings(
+                threshold=threshold,
+                warmup=warmup,
+                h_symmetry_time_pct=h_symmetry_time_pct,
+                v_symmetry_time_pct=v_symmetry_time_pct,
+            ),
+        ).add_scheduler_args_if_applicable(pipeline.scheduler, eta=ddim_eta)
+
+        def make_image(x_T: torch.Tensor, seed: int):
+            # FIXME: use x_T for initial seeded noise
+            # We're not at the moment because the pipeline automatically resizes init_image if
+            # necessary, which the x_T input might not match.
+            # In the meantime, reset the seed prior to generating pipeline output so we at least get the same result.
+            logging.set_verbosity_error()  # quench safety check warnings
+            pipeline_output = pipeline.img2img_from_embeddings(
+                init_image,
+                strength,
+                steps,
+                conditioning_data,
+                noise_func=self.get_noise_like,
+                callback=step_callback,
+                seed=seed,
+            )
+            if (
+                pipeline_output.attention_map_saver is not None
+                and attention_maps_callback is not None
+            ):
+                attention_maps_callback(pipeline_output.attention_map_saver)
+            return pipeline.numpy_to_pil(pipeline_output.images)[0]
+
+        return make_image
+
+    def get_noise_like(self, like: torch.Tensor):
+        device = like.device
+        if device.type == "mps":
+            x = torch.randn_like(like, device="cpu").to(device)
+        else:
+            x = torch.randn_like(like, device=device)
+        if self.perlin > 0.0:
+            shape = like.shape
+            x = (1 - self.perlin) * x + self.perlin * self.get_perlin_noise(
+                shape[3], shape[2]
+            )
+        return x

invokeai/backend/generator/inpaint.py

@@ -1,32 +1,35 @@
-'''
-ldm.invoke.generator.inpaint descends from ldm.invoke.generator
-'''
+"""
+invokeai.backend.generator.inpaint descends from .generator
+"""
 from __future__ import annotations
 
 import math
 
-import PIL
 import cv2
 import numpy as np
+import PIL
 import torch
-from PIL import Image, ImageFilter, ImageOps, ImageChops
+from PIL import Image, ImageChops, ImageFilter, ImageOps
 
-from ldm.invoke.generator.diffusers_pipeline import image_resized_to_grid_as_tensor, StableDiffusionGeneratorPipeline, \
-    ConditioningData
-from ldm.invoke.generator.img2img import Img2Img
-from ldm.invoke.patchmatch import PatchMatch
-from ldm.util import debug_image
+from ..image_util import PatchMatch, debug_image
+from ..stable_diffusion.diffusers_pipeline import (
+    ConditioningData,
+    StableDiffusionGeneratorPipeline,
+    image_resized_to_grid_as_tensor,
+)
+from .img2img import Img2Img
 
 
-def infill_methods()->list[str]:
+def infill_methods() -> list[str]:
     methods = [
         "tile",
         "solid",
     ]
     if PatchMatch.patchmatch_available():
-        methods.insert(0, 'patchmatch')
+        methods.insert(0, "patchmatch")
     return methods
 
+
 class Inpaint(Img2Img):
     def __init__(self, model, precision):
         self.inpaint_height = 0
@@ -53,11 +56,11 @@ class Inpaint(Img2Img):
             np.ravel(image),
             shape=(nrows, ncols, height, width, depth),
             strides=(height * _strides[0], width * _strides[1], *_strides),
-            writeable=False
+            writeable=False,
         )
 
     def infill_patchmatch(self, im: Image.Image) -> Image:
-        if im.mode != 'RGBA':
+        if im.mode != "RGBA":
             return im
 
         # Skip patchmatch if patchmatch isn't available
@@ -65,13 +68,17 @@ class Inpaint(Img2Img):
             return im
 
         # Patchmatch (note, we may want to expose patch_size? Increasing it significantly impacts performance though)
-        im_patched_np = PatchMatch.inpaint(im.convert('RGB'), ImageOps.invert(im.split()[-1]), patch_size = 3)
-        im_patched = Image.fromarray(im_patched_np, mode = 'RGB')
+        im_patched_np = PatchMatch.inpaint(
+            im.convert("RGB"), ImageOps.invert(im.split()[-1]), patch_size=3
+        )
+        im_patched = Image.fromarray(im_patched_np, mode="RGB")
         return im_patched
 
-    def tile_fill_missing(self, im: Image.Image, tile_size: int = 16, seed: int = None) -> Image:
+    def tile_fill_missing(
+        self, im: Image.Image, tile_size: int = 16, seed: int = None
+    ) -> Image:
         # Only fill if there's an alpha layer
-        if im.mode != 'RGBA':
+        if im.mode != "RGBA":
             return im
 
         a = np.asarray(im, dtype=np.uint8)
@@ -79,21 +86,21 @@ class Inpaint(Img2Img):
         tile_size = (tile_size, tile_size)
 
         # Get the image as tiles of a specified size
-        tiles = self.get_tile_images(a,*tile_size).copy()
+        tiles = self.get_tile_images(a, *tile_size).copy()
 
         # Get the mask as tiles
-        tiles_mask = tiles[:,:,:,:,3]
+        tiles_mask = tiles[:, :, :, :, 3]
 
         # Find any mask tiles with any fully transparent pixels (we will be replacing these later)
         tmask_shape = tiles_mask.shape
         tiles_mask = tiles_mask.reshape(math.prod(tiles_mask.shape))
-        n,ny = (math.prod(tmask_shape[0:2])), math.prod(tmask_shape[2:])
-        tiles_mask = (tiles_mask > 0)
-        tiles_mask = tiles_mask.reshape((n,ny)).all(axis = 1)
+        n, ny = (math.prod(tmask_shape[0:2])), math.prod(tmask_shape[2:])
+        tiles_mask = tiles_mask > 0
+        tiles_mask = tiles_mask.reshape((n, ny)).all(axis=1)
 
         # Get RGB tiles in single array and filter by the mask
         tshape = tiles.shape
-        tiles_all = tiles.reshape((math.prod(tiles.shape[0:2]), * tiles.shape[2:]))
+        tiles_all = tiles.reshape((math.prod(tiles.shape[0:2]), *tiles.shape[2:]))
         filtered_tiles = tiles_all[tiles_mask]
 
         if len(filtered_tiles) == 0:
@@ -101,23 +108,32 @@ class Inpaint(Img2Img):
 
         # Find all invalid tiles and replace with a random valid tile
         replace_count = (tiles_mask == False).sum()
-        rng = np.random.default_rng(seed = seed)
-        tiles_all[np.logical_not(tiles_mask)] = filtered_tiles[rng.choice(filtered_tiles.shape[0], replace_count),:,:,:]
+        rng = np.random.default_rng(seed=seed)
+        tiles_all[np.logical_not(tiles_mask)] = filtered_tiles[
+            rng.choice(filtered_tiles.shape[0], replace_count), :, :, :
+        ]
 
         # Convert back to an image
         tiles_all = tiles_all.reshape(tshape)
-        tiles_all = tiles_all.swapaxes(1,2)
-        st = tiles_all.reshape((math.prod(tiles_all.shape[0:2]), math.prod(tiles_all.shape[2:4]), tiles_all.shape[4]))
-        si = Image.fromarray(st, mode='RGBA')
+        tiles_all = tiles_all.swapaxes(1, 2)
+        st = tiles_all.reshape(
+            (
+                math.prod(tiles_all.shape[0:2]),
+                math.prod(tiles_all.shape[2:4]),
+                tiles_all.shape[4],
+            )
+        )
+        si = Image.fromarray(st, mode="RGBA")
 
         return si
 
-
     def mask_edge(self, mask: Image, edge_size: int, edge_blur: int) -> Image:
         npimg = np.asarray(mask, dtype=np.uint8)
 
         # Detect any partially transparent regions
-        npgradient = np.uint8(255 * (1.0 - np.floor(np.abs(0.5 - np.float32(npimg) / 255.0) * 2.0)))
+        npgradient = np.uint8(
+            255 * (1.0 - np.floor(np.abs(0.5 - np.float32(npimg) / 255.0) * 2.0))
+        )
 
         # Detect hard edges
         npedge = cv2.Canny(npimg, threshold1=100, threshold2=200)
@@ -126,7 +142,9 @@ class Inpaint(Img2Img):
         npmask = npgradient + npedge
 
         # Expand
-        npmask = cv2.dilate(npmask, np.ones((3,3), np.uint8), iterations = int(edge_size / 2))
+        npmask = cv2.dilate(
+            npmask, np.ones((3, 3), np.uint8), iterations=int(edge_size / 2)
+        )
 
         new_mask = Image.fromarray(npmask)
 
@@ -135,9 +153,23 @@ class Inpaint(Img2Img):
 
         return ImageOps.invert(new_mask)
 
-
-    def seam_paint(self, im: Image.Image, seam_size: int, seam_blur: int, prompt, sampler, steps, cfg_scale, ddim_eta,
-                   conditioning, strength, noise, infill_method, step_callback) -> Image.Image:
+    def seam_paint(
+        self,
+        im: Image.Image,
+        seam_size: int,
+        seam_blur: int,
+        prompt,
+        seed,
+        sampler,
+        steps,
+        cfg_scale,
+        ddim_eta,
+        conditioning,
+        strength,
+        noise,
+        infill_method,
+        step_callback,
+    ) -> Image.Image:
         hard_mask = self.pil_image.split()[-1].copy()
         mask = self.mask_edge(hard_mask, seam_size, seam_blur)
 
@@ -148,45 +180,52 @@ class Inpaint(Img2Img):
             cfg_scale,
             ddim_eta,
             conditioning,
-            init_image = im.copy().convert('RGBA'),
-            mask_image = mask,
-            strength = strength,
-            mask_blur_radius = 0,
-            seam_size = 0,
-            step_callback = step_callback,
-            inpaint_width = im.width,
-            inpaint_height = im.height,
-            infill_method = infill_method
+            init_image=im.copy().convert("RGBA"),
+            mask_image=mask,
+            strength=strength,
+            mask_blur_radius=0,
+            seam_size=0,
+            step_callback=step_callback,
+            inpaint_width=im.width,
+            inpaint_height=im.height,
+            infill_method=infill_method,
         )
 
         seam_noise = self.get_noise(im.width, im.height)
 
-        result = make_image(seam_noise)
+        result = make_image(seam_noise, seed)
 
         return result
 
-
     @torch.no_grad()
-    def get_make_image(self,prompt,sampler,steps,cfg_scale,ddim_eta,
-                       conditioning,
-                       init_image: PIL.Image.Image | torch.FloatTensor,
-                       mask_image: PIL.Image.Image | torch.FloatTensor,
-                       strength: float,
-                       mask_blur_radius: int = 8,
-                       # Seam settings - when 0, doesn't fill seam
-                       seam_size: int = 0,
-                       seam_blur: int = 0,
-                       seam_strength: float = 0.7,
-                       seam_steps: int = 10,
-                       tile_size: int = 32,
-                       step_callback=None,
-                       inpaint_replace=False, enable_image_debugging=False,
-                       infill_method = None,
-                       inpaint_width=None,
-                       inpaint_height=None,
-                       inpaint_fill:tuple(int)=(0x7F, 0x7F, 0x7F, 0xFF),
-                       attention_maps_callback=None,
-                       **kwargs):
+    def get_make_image(
+        self,
+        prompt,
+        sampler,
+        steps,
+        cfg_scale,
+        ddim_eta,
+        conditioning,
+        init_image: PIL.Image.Image | torch.FloatTensor,
+        mask_image: PIL.Image.Image | torch.FloatTensor,
+        strength: float,
+        mask_blur_radius: int = 8,
+        # Seam settings - when 0, doesn't fill seam
+        seam_size: int = 0,
+        seam_blur: int = 0,
+        seam_strength: float = 0.7,
+        seam_steps: int = 10,
+        tile_size: int = 32,
+        step_callback=None,
+        inpaint_replace=False,
+        enable_image_debugging=False,
+        infill_method=None,
+        inpaint_width=None,
+        inpaint_height=None,
+        inpaint_fill: tuple(int) = (0x7F, 0x7F, 0x7F, 0xFF),
+        attention_maps_callback=None,
+        **kwargs,
+    ):
         """
         Returns a function returning an image derived from the prompt and
         the initial image + mask.  Return value depends on the seed at
@@ -204,33 +243,39 @@ class Inpaint(Img2Img):
             self.pil_image = init_image.copy()
 
             # Do infill
-            if infill_method == 'patchmatch' and PatchMatch.patchmatch_available():
+            if infill_method == "patchmatch" and PatchMatch.patchmatch_available():
                 init_filled = self.infill_patchmatch(self.pil_image.copy())
-            elif infill_method == 'tile':
+            elif infill_method == "tile":
                 init_filled = self.tile_fill_missing(
-                    self.pil_image.copy(),
-                    seed = self.seed,
-                    tile_size = tile_size
+                    self.pil_image.copy(), seed=self.seed, tile_size=tile_size
                 )
-            elif infill_method == 'solid':
+            elif infill_method == "solid":
                 solid_bg = PIL.Image.new("RGBA", init_image.size, inpaint_fill)
                 init_filled = PIL.Image.alpha_composite(solid_bg, init_image)
             else:
-                raise ValueError(f"Non-supported infill type {infill_method}", infill_method)
-            init_filled.paste(init_image, (0,0), init_image.split()[-1])
+                raise ValueError(
+                    f"Non-supported infill type {infill_method}", infill_method
+                )
+            init_filled.paste(init_image, (0, 0), init_image.split()[-1])
 
             # Resize if requested for inpainting
             if inpaint_width and inpaint_height:
                 init_filled = init_filled.resize((inpaint_width, inpaint_height))
 
-            debug_image(init_filled, "init_filled", debug_status=self.enable_image_debugging)
+            debug_image(
+                init_filled, "init_filled", debug_status=self.enable_image_debugging
+            )
 
             # Create init tensor
-            init_image = image_resized_to_grid_as_tensor(init_filled.convert('RGB'))
+            init_image = image_resized_to_grid_as_tensor(init_filled.convert("RGB"))
 
         if isinstance(mask_image, PIL.Image.Image):
             self.pil_mask = mask_image.copy()
-            debug_image(mask_image, "mask_image BEFORE multiply with pil_image", debug_status=self.enable_image_debugging)
+            debug_image(
+                mask_image,
+                "mask_image BEFORE multiply with pil_image",
+                debug_status=self.enable_image_debugging,
+            )
 
             init_alpha = self.pil_image.getchannel("A")
             if mask_image.mode != "L":
@@ -243,8 +288,14 @@ class Inpaint(Img2Img):
             if inpaint_width and inpaint_height:
                 mask_image = mask_image.resize((inpaint_width, inpaint_height))
 
-            debug_image(mask_image, "mask_image AFTER multiply with pil_image", debug_status=self.enable_image_debugging)
-            mask: torch.FloatTensor = image_resized_to_grid_as_tensor(mask_image, normalize=False)
+            debug_image(
+                mask_image,
+                "mask_image AFTER multiply with pil_image",
+                debug_status=self.enable_image_debugging,
+            )
+            mask: torch.FloatTensor = image_resized_to_grid_as_tensor(
+                mask_image, normalize=False
+            )
         else:
             mask: torch.FloatTensor = mask_image
 
@@ -256,11 +307,11 @@ class Inpaint(Img2Img):
 
         # todo: support cross-attention control
         uc, c, _ = conditioning
-        conditioning_data = (ConditioningData(uc, c, cfg_scale)
-                             .add_scheduler_args_if_applicable(pipeline.scheduler, eta=ddim_eta))
+        conditioning_data = ConditioningData(
+            uc, c, cfg_scale
+        ).add_scheduler_args_if_applicable(pipeline.scheduler, eta=ddim_eta)
 
-
-        def make_image(x_T):
+        def make_image(x_T: torch.Tensor, seed: int):
             pipeline_output = pipeline.inpaint_from_embeddings(
                 init_image=init_image,
                 mask=1 - mask,  # expects white means "paint here."
@@ -269,45 +320,75 @@ class Inpaint(Img2Img):
                 conditioning_data=conditioning_data,
                 noise_func=self.get_noise_like,
                 callback=step_callback,
+                seed=seed,
             )
 
-            if pipeline_output.attention_map_saver is not None and attention_maps_callback is not None:
+            if (
+                pipeline_output.attention_map_saver is not None
+                and attention_maps_callback is not None
+            ):
                 attention_maps_callback(pipeline_output.attention_map_saver)
 
-            result = self.postprocess_size_and_mask(pipeline.numpy_to_pil(pipeline_output.images)[0])
+            result = self.postprocess_size_and_mask(
+                pipeline.numpy_to_pil(pipeline_output.images)[0]
+            )
 
             # Seam paint if this is our first pass (seam_size set to 0 during seam painting)
             if seam_size > 0:
                 old_image = self.pil_image or init_image
                 old_mask = self.pil_mask or mask_image
 
-                result = self.seam_paint(result, seam_size, seam_blur, prompt, sampler, seam_steps, cfg_scale, ddim_eta,
-                                         conditioning, seam_strength, x_T, infill_method, step_callback)
+                result = self.seam_paint(
+                    result,
+                    seam_size,
+                    seam_blur,
+                    prompt,
+                    seed,
+                    sampler,
+                    seam_steps,
+                    cfg_scale,
+                    ddim_eta,
+                    conditioning,
+                    seam_strength,
+                    x_T,
+                    infill_method,
+                    step_callback,
+                )
 
                 # Restore original settings
-                self.get_make_image(prompt,sampler,steps,cfg_scale,ddim_eta,
-                       conditioning,
-                       old_image,
-                       old_mask,
-                       strength,
-                       mask_blur_radius, seam_size, seam_blur, seam_strength,
-                       seam_steps, tile_size, step_callback,
-                       inpaint_replace, enable_image_debugging,
-                       inpaint_width = inpaint_width,
-                       inpaint_height = inpaint_height,
-                       infill_method = infill_method,
-                       **kwargs)
+                self.get_make_image(
+                    prompt,
+                    sampler,
+                    steps,
+                    cfg_scale,
+                    ddim_eta,
+                    conditioning,
+                    old_image,
+                    old_mask,
+                    strength,
+                    mask_blur_radius,
+                    seam_size,
+                    seam_blur,
+                    seam_strength,
+                    seam_steps,
+                    tile_size,
+                    step_callback,
+                    inpaint_replace,
+                    enable_image_debugging,
+                    inpaint_width=inpaint_width,
+                    inpaint_height=inpaint_height,
+                    infill_method=infill_method,
+                    **kwargs,
+                )
 
             return result
 
         return make_image
 
-
-    def sample_to_image(self, samples)->Image.Image:
-        gen_result = super().sample_to_image(samples).convert('RGB')
+    def sample_to_image(self, samples) -> Image.Image:
+        gen_result = super().sample_to_image(samples).convert("RGB")
         return self.postprocess_size_and_mask(gen_result)
 
-
     def postprocess_size_and_mask(self, gen_result: Image.Image) -> Image.Image:
         debug_image(gen_result, "gen_result", debug_status=self.enable_image_debugging)
 
@@ -318,7 +399,13 @@ class Inpaint(Img2Img):
         if self.pil_image is None or self.pil_mask is None:
             return gen_result
 
-        corrected_result = self.repaste_and_color_correct(gen_result, self.pil_image, self.pil_mask, self.mask_blur_radius)
-        debug_image(corrected_result, "corrected_result", debug_status=self.enable_image_debugging)
+        corrected_result = self.repaste_and_color_correct(
+            gen_result, self.pil_image, self.pil_mask, self.mask_blur_radius
+        )
+        debug_image(
+            corrected_result,
+            "corrected_result",
+            debug_status=self.enable_image_debugging,
+        )
 
         return corrected_result

invokeai/backend/generator/txt2img.py

@@ -0,0 +1,81 @@
+"""
+invokeai.backend.generator.txt2img inherits from invokeai.backend.generator
+"""
+import PIL.Image
+import torch
+
+from ..stable_diffusion import (
+    ConditioningData,
+    PostprocessingSettings,
+    StableDiffusionGeneratorPipeline,
+)
+from .base import Generator
+
+
+class Txt2Img(Generator):
+    def __init__(self, model, precision):
+        super().__init__(model, precision)
+
+    @torch.no_grad()
+    def get_make_image(
+        self,
+        prompt,
+        sampler,
+        steps,
+        cfg_scale,
+        ddim_eta,
+        conditioning,
+        width,
+        height,
+        step_callback=None,
+        threshold=0.0,
+        warmup=0.2,
+        perlin=0.0,
+        h_symmetry_time_pct=None,
+        v_symmetry_time_pct=None,
+        attention_maps_callback=None,
+        **kwargs,
+    ):
+        """
+        Returns a function returning an image derived from the prompt and the initial image
+        Return value depends on the seed at the time you call it
+        kwargs are 'width' and 'height'
+        """
+        self.perlin = perlin
+
+        # noinspection PyTypeChecker
+        pipeline: StableDiffusionGeneratorPipeline = self.model
+        pipeline.scheduler = sampler
+
+        uc, c, extra_conditioning_info = conditioning
+        conditioning_data = ConditioningData(
+            uc,
+            c,
+            cfg_scale,
+            extra_conditioning_info,
+            postprocessing_settings=PostprocessingSettings(
+                threshold=threshold,
+                warmup=warmup,
+                h_symmetry_time_pct=h_symmetry_time_pct,
+                v_symmetry_time_pct=v_symmetry_time_pct,
+            ),
+        ).add_scheduler_args_if_applicable(pipeline.scheduler, eta=ddim_eta)
+
+        def make_image(x_T: torch.Tensor, _: int) -> PIL.Image.Image:
+            pipeline_output = pipeline.image_from_embeddings(
+                latents=torch.zeros_like(x_T, dtype=self.torch_dtype()),
+                noise=x_T,
+                num_inference_steps=steps,
+                conditioning_data=conditioning_data,
+                callback=step_callback,
+            )
+
+            if (
+                pipeline_output.attention_map_saver is not None
+                and attention_maps_callback is not None
+            ):
+                attention_maps_callback(pipeline_output.attention_map_saver)
+
+            return pipeline.numpy_to_pil(pipeline_output.images)[0]
+
+        return make_image