(ci) remove python 3.10 from the test matrix; comment out GPU tests for now

.dockerignore

@@ -1,11 +1,9 @@
 *
 !invokeai
 !pyproject.toml
-!uv.lock
 !docker/docker-entrypoint.sh
 !LICENSE
 
-**/dist
 **/node_modules
 **/__pycache__
-**/*.egg-info
+**/*.egg-info

.github/workflows/build-container.yml

@@ -97,8 +97,6 @@ jobs:
           context: .
           file: docker/Dockerfile
           platforms: ${{ env.PLATFORMS }}
-          build-args: |
-            GPU_DRIVER=${{ matrix.gpu-driver }}
           push: ${{ github.ref == 'refs/heads/main' || github.ref_type == 'tag' || github.event.inputs.push-to-registry }}
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}

.github/workflows/python-checks.yml

@@ -34,9 +34,6 @@ on:
 
 jobs:
   python-checks:
-    env:
-      # uv requires a venv by default - but for this, we can simply use the system python
-      UV_SYSTEM_PYTHON: 1
     runs-on: ubuntu-latest
     timeout-minutes: 5 # expected run time: <1 min
     steps:
@@ -60,19 +57,25 @@ jobs:
               - '!invokeai/frontend/web/**'
               - 'tests/**'
 
-      - name: setup uv
+      - name: setup python
         if: ${{ steps.changed-files.outputs.python_any_changed == 'true' || inputs.always_run == true }}
-        uses: astral-sh/setup-uv@v5
+        uses: actions/setup-python@v5
         with:
-          version: '0.6.10'
-          enable-cache: true
+          python-version: '3.12'
+          cache: pip
+          cache-dependency-path: pyproject.toml
+
+      - name: install ruff
+        if: ${{ steps.changed-files.outputs.python_any_changed == 'true' || inputs.always_run == true }}
+        run: pip install ruff==0.11.2
+        shell: bash
 
       - name: ruff check
         if: ${{ steps.changed-files.outputs.python_any_changed == 'true' || inputs.always_run == true }}
-        run: uv tool run ruff@0.11.2 check --output-format=github .
+        run: ruff check --output-format=github .
         shell: bash
 
       - name: ruff format
         if: ${{ steps.changed-files.outputs.python_any_changed == 'true' || inputs.always_run == true }}
-        run: uv tool run ruff@0.11.2 format --check .
+        run: ruff format --check .
         shell: bash

.github/workflows/python-tests.yml

@@ -40,16 +40,24 @@ jobs:
       matrix:
         python-version:
           - '3.11'
-          - '3.12'
         platform:
+          # - linux-cuda-12_6
+          # - linux-rocm-6_2
           - linux-cpu
           - macos-default
           - windows-cpu
         include:
+          # - platform: linux-cuda-12_6
+          #   os: ubuntu-24.04
+          #   github-env: $GITHUB_ENV
+          # - platform: linux-rocm-6_2
+          #   os: ubuntu-24.04
+          #   extra-index-url: 'https://download.pytorch.org/whl/rocm6.2'
+          #   github-env: $GITHUB_ENV
           - platform: linux-cpu
             os: ubuntu-24.04
-            extra-index-url: 'https://download.pytorch.org/whl/cpu'
             github-env: $GITHUB_ENV
+            extra-index-url: 'https://download.pytorch.org/whl/cpu'
           - platform: macos-default
             os: macOS-14
             github-env: $GITHUB_ENV
@@ -61,8 +69,6 @@ jobs:
     timeout-minutes: 15 # expected run time: 2-6 min, depending on platform
     env:
       PIP_USE_PEP517: '1'
-      UV_SYSTEM_PYTHON: 1
-
     steps:
       - name: checkout
         # https://github.com/nschloe/action-cached-lfs-checkout
@@ -85,25 +91,20 @@ jobs:
               - '!invokeai/frontend/web/**'
               - 'tests/**'
 
-      - name: setup uv
-        if: ${{ steps.changed-files.outputs.python_any_changed == 'true' || inputs.always_run == true }}
-        uses: astral-sh/setup-uv@v5
-        with:
-          version: '0.6.10'
-          enable-cache: true
-          python-version: ${{ matrix.python-version }}
-
       - name: setup python
         if: ${{ steps.changed-files.outputs.python_any_changed == 'true' || inputs.always_run == true }}
         uses: actions/setup-python@v5
         with:
           python-version: ${{ matrix.python-version }}
+          cache: pip
+          cache-dependency-path: pyproject.toml
 
       - name: install dependencies
         if: ${{ steps.changed-files.outputs.python_any_changed == 'true' || inputs.always_run == true }}
         env:
-          UV_INDEX: ${{ matrix.extra-index-url }}
-        run: uv pip install --editable ".[test]"
+          PIP_EXTRA_INDEX_URL: ${{ matrix.extra-index-url }}
+        run: >
+          pip3 install --editable=".[test]"
 
       - name: run pytest
         if: ${{ steps.changed-files.outputs.python_any_changed == 'true' || inputs.always_run == true }}

.github/workflows/typegen-checks.yml

@@ -54,25 +54,17 @@ jobs:
               - 'pyproject.toml'
               - 'invokeai/**'
 
-      - name: setup uv
-        if: ${{ steps.changed-files.outputs.src_any_changed == 'true' || inputs.always_run == true }}
-        uses: astral-sh/setup-uv@v5
-        with:
-          version: '0.6.10'
-          enable-cache: true
-          python-version: '3.11'
-
       - name: setup python
         if: ${{ steps.changed-files.outputs.src_any_changed == 'true' || inputs.always_run == true }}
         uses: actions/setup-python@v5
         with:
-          python-version: '3.11'
+          python-version: '3.10'
+          cache: pip
+          cache-dependency-path: pyproject.toml
 
-      - name: install dependencies
+      - name: install python dependencies
         if: ${{ steps.changed-files.outputs.src_any_changed == 'true' || inputs.always_run == true }}
-        env:
-          UV_INDEX: ${{ matrix.extra-index-url }}
-        run: uv pip install --editable .
+        run: pip3 install --use-pep517 --editable="."
 
       - name: install frontend dependencies
         if: ${{ steps.changed-files.outputs.src_any_changed == 'true' || inputs.always_run == true }}
@@ -85,7 +77,7 @@ jobs:
 
       - name: generate schema
         if: ${{ steps.changed-files.outputs.src_any_changed == 'true' || inputs.always_run == true }}
-        run: cd invokeai/frontend/web && uv run ../../../scripts/generate_openapi_schema.py | pnpm typegen
+        run: make frontend-typegen
         shell: bash
 
       - name: compare files

.nvmrc

@@ -1 +1 @@
-v22.14.0
+v22.12.0

docker/Dockerfile

@@ -1,6 +1,77 @@
 # syntax=docker/dockerfile:1.4
 
-#### Web UI ------------------------------------
+## Builder stage
+
+FROM library/ubuntu:24.04 AS builder
+
+ARG DEBIAN_FRONTEND=noninteractive
+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
+RUN --mount=type=cache,target=/var/cache/apt,sharing=locked \
+    --mount=type=cache,target=/var/lib/apt,sharing=locked \
+    apt update && apt-get install -y \
+        build-essential \
+        git
+
+# Install `uv` for package management
+COPY --from=ghcr.io/astral-sh/uv:0.6.0 /uv /uvx /bin/
+
+ENV VIRTUAL_ENV=/opt/venv
+ENV PATH="$VIRTUAL_ENV/bin:$PATH"
+ENV INVOKEAI_SRC=/opt/invokeai
+ENV PYTHON_VERSION=3.11
+ENV UV_PYTHON=3.11
+ENV UV_COMPILE_BYTECODE=1
+ENV UV_LINK_MODE=copy
+ENV UV_PROJECT_ENVIRONMENT="$VIRTUAL_ENV"
+ENV UV_INDEX="https://download.pytorch.org/whl/cu124"
+
+ARG GPU_DRIVER=cuda
+# unused but available
+ARG BUILDPLATFORM
+
+# Switch to the `ubuntu` user to work around dependency issues with uv-installed python
+RUN mkdir -p ${VIRTUAL_ENV} && \
+    mkdir -p ${INVOKEAI_SRC} && \
+    chmod -R a+w /opt && \
+    mkdir ~ubuntu/.cache && chown ubuntu: ~ubuntu/.cache
+USER ubuntu
+
+# Install python
+RUN --mount=type=cache,target=/home/ubuntu/.cache/uv,uid=1000,gid=1000 \
+  uv python install ${PYTHON_VERSION}
+
+WORKDIR ${INVOKEAI_SRC}
+
+# Install project's dependencies as a separate layer so they aren't rebuilt every commit.
+# bind-mount instead of copy to defer adding sources to the image until next layer.
+#
+# NOTE: there are no pytorch builds for arm64 + cuda, only cpu
+# x86_64/CUDA is the default
+RUN --mount=type=cache,target=/home/ubuntu/.cache/uv,uid=1000,gid=1000 \
+    --mount=type=bind,source=pyproject.toml,target=pyproject.toml \
+    --mount=type=bind,source=invokeai/version,target=invokeai/version \
+    if [ "$TARGETPLATFORM" = "linux/arm64" ] || [ "$GPU_DRIVER" = "cpu" ]; then \
+      UV_INDEX="https://download.pytorch.org/whl/cpu"; \
+    elif [ "$GPU_DRIVER" = "rocm" ]; then \
+      UV_INDEX="https://download.pytorch.org/whl/rocm6.1"; \
+    fi && \
+    uv sync --no-install-project
+
+# Now that the bulk of the dependencies have been installed, copy in the project files that change more frequently.
+COPY invokeai invokeai
+COPY pyproject.toml .
+
+RUN --mount=type=cache,target=/home/ubuntu/.cache/uv,uid=1000,gid=1000 \
+    --mount=type=bind,source=pyproject.toml,target=pyproject.toml \
+    if [ "$TARGETPLATFORM" = "linux/arm64" ] || [ "$GPU_DRIVER" = "cpu" ]; then \
+      UV_INDEX="https://download.pytorch.org/whl/cpu"; \
+    elif [ "$GPU_DRIVER" = "rocm" ]; then \
+      UV_INDEX="https://download.pytorch.org/whl/rocm6.1"; \
+    fi && \
+    uv sync
+
+
+#### Build the Web UI ------------------------------------
 
 FROM docker.io/node:22-slim AS web-builder
 ENV PNPM_HOME="/pnpm"
@@ -14,89 +85,69 @@ RUN --mount=type=cache,target=/pnpm/store \
     pnpm install --frozen-lockfile
 RUN npx vite build
 
-## Backend ---------------------------------------
+#### Runtime stage ---------------------------------------
 
-FROM library/ubuntu:24.04
+FROM library/ubuntu:24.04 AS runtime
 
 ARG DEBIAN_FRONTEND=noninteractive
-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
-RUN --mount=type=cache,target=/var/cache/apt \
-    --mount=type=cache,target=/var/lib/apt \
-    apt update && apt install -y --no-install-recommends \
-    ca-certificates \
-    git \
-    gosu \
-    libglib2.0-0 \
-    libgl1 \
-    libglx-mesa0 \
-    build-essential \
-    libopencv-dev \
-    libstdc++-10-dev
+ENV PYTHONUNBUFFERED=1
+ENV PYTHONDONTWRITEBYTECODE=1
 
-ENV \
-    PYTHONUNBUFFERED=1 \
-    PYTHONDONTWRITEBYTECODE=1 \
-    VIRTUAL_ENV=/opt/venv \
-    INVOKEAI_SRC=/opt/invokeai \
-    PYTHON_VERSION=3.12 \
-    UV_PYTHON=3.12 \
-    UV_COMPILE_BYTECODE=1 \
-    UV_MANAGED_PYTHON=1 \
-    UV_LINK_MODE=copy \
-    UV_PROJECT_ENVIRONMENT=/opt/venv \
-    UV_INDEX="https://download.pytorch.org/whl/cu124" \
-    INVOKEAI_ROOT=/invokeai \
-    INVOKEAI_HOST=0.0.0.0 \
-    INVOKEAI_PORT=9090 \
-    PATH="/opt/venv/bin:$PATH" \
-    CONTAINER_UID=${CONTAINER_UID:-1000} \
-    CONTAINER_GID=${CONTAINER_GID:-1000}
+RUN apt update && apt install -y --no-install-recommends \
+        git \
+        curl \
+        vim \
+        tmux \
+        ncdu \
+        iotop \
+        bzip2 \
+        gosu \
+        magic-wormhole \
+        libglib2.0-0 \
+        libgl1 \
+        libglx-mesa0 \
+        build-essential \
+        libopencv-dev \
+        libstdc++-10-dev &&\
+    apt-get clean && apt-get autoclean
 
-ARG GPU_DRIVER=cuda
+ENV INVOKEAI_SRC=/opt/invokeai
+ENV VIRTUAL_ENV=/opt/venv
+ENV UV_PROJECT_ENVIRONMENT="$VIRTUAL_ENV"
+ENV PYTHON_VERSION=3.11
+ENV INVOKEAI_ROOT=/invokeai
+ENV INVOKEAI_HOST=0.0.0.0
+ENV INVOKEAI_PORT=9090
+ENV PATH="$VIRTUAL_ENV/bin:$INVOKEAI_SRC:$PATH"
+ENV CONTAINER_UID=${CONTAINER_UID:-1000}
+ENV CONTAINER_GID=${CONTAINER_GID:-1000}
 
 # Install `uv` for package management
-COPY --from=ghcr.io/astral-sh/uv:0.6.9 /uv /uvx /bin/
+# and install python for the ubuntu user (expected to exist on ubuntu >=24.x)
+# this is too tiny to optimize with multi-stage builds, but maybe we'll come back to it
+COPY --from=ghcr.io/astral-sh/uv:0.6.0 /uv /uvx /bin/
+USER ubuntu
+RUN uv python install ${PYTHON_VERSION}
+USER root
 
-# Install python & allow non-root user to use it by traversing the /root dir without read permissions
-RUN --mount=type=cache,target=/root/.cache/uv \
-    uv python install ${PYTHON_VERSION} && \
-    # chmod --recursive a+rX /root/.local/share/uv/python
-    chmod 711 /root
-
-WORKDIR ${INVOKEAI_SRC}
-
-# Install project's dependencies as a separate layer so they aren't rebuilt every commit.
-# bind-mount instead of copy to defer adding sources to the image until next layer.
-#
-# NOTE: there are no pytorch builds for arm64 + cuda, only cpu
-# x86_64/CUDA is the default
-RUN --mount=type=cache,target=/root/.cache/uv \
-    --mount=type=bind,source=pyproject.toml,target=pyproject.toml \
-    --mount=type=bind,source=uv.lock,target=uv.lock \
-    # this is just to get the package manager to recognize that the project exists, without making changes to the docker layer
-    --mount=type=bind,source=invokeai/version,target=invokeai/version \
-    if [ "$TARGETPLATFORM" = "linux/arm64" ] || [ "$GPU_DRIVER" = "cpu" ]; then UV_INDEX="https://download.pytorch.org/whl/cpu"; \
-    elif [ "$GPU_DRIVER" = "rocm" ]; then UV_INDEX="https://download.pytorch.org/whl/rocm6.2"; \
-    fi && \
-    uv sync --frozen
-
-# build patchmatch
-RUN cd /usr/lib/$(uname -p)-linux-gnu/pkgconfig/ && ln -sf opencv4.pc opencv.pc
-RUN python -c "from patchmatch import patch_match"
+# --link requires buldkit w/ dockerfile syntax 1.4
+COPY --link --from=builder ${INVOKEAI_SRC} ${INVOKEAI_SRC}
+COPY --link --from=builder ${VIRTUAL_ENV} ${VIRTUAL_ENV}
+COPY --link --from=web-builder /build/dist ${INVOKEAI_SRC}/invokeai/frontend/web/dist
 
 # Link amdgpu.ids for ROCm builds
 # contributed by https://github.com/Rubonnek
 RUN mkdir -p "/opt/amdgpu/share/libdrm" &&\
-    ln -s "/usr/share/libdrm/amdgpu.ids" "/opt/amdgpu/share/libdrm/amdgpu.ids"
+  ln -s "/usr/share/libdrm/amdgpu.ids" "/opt/amdgpu/share/libdrm/amdgpu.ids"
+
+WORKDIR ${INVOKEAI_SRC}
+
+# build patchmatch
+RUN cd /usr/lib/$(uname -p)-linux-gnu/pkgconfig/ && ln -sf opencv4.pc opencv.pc
+RUN python -c "from patchmatch import patch_match"
 
 RUN mkdir -p ${INVOKEAI_ROOT} && chown -R ${CONTAINER_UID}:${CONTAINER_GID} ${INVOKEAI_ROOT}
 
 COPY docker/docker-entrypoint.sh ./
 ENTRYPOINT ["/opt/invokeai/docker-entrypoint.sh"]
 CMD ["invokeai-web"]
-
-# --link requires buldkit w/ dockerfile syntax 1.4, does not work with podman
-COPY --link --from=web-builder /build/dist ${INVOKEAI_SRC}/invokeai/frontend/web/dist
-
-# add sources last to minimize image changes on code changes
-COPY invokeai ${INVOKEAI_SRC}/invokeai

docs/contributing/dev-environment.md

@@ -41,7 +41,7 @@ If you just want to use Invoke, you should use the [launcher][launcher link].
      With the modifications made, the install command should look something like this:
 
       ```sh
-      uv pip install -e ".[dev,test,docs,xformers]" --python 3.12 --python-preference only-managed --index=https://download.pytorch.org/whl/cu124 --reinstall
+      uv pip install -e ".[dev,test,docs,xformers]" --python 3.11 --python-preference only-managed --index=https://download.pytorch.org/whl/cu124 --reinstall
       ```
 
 6. At this point, you should have Invoke installed, a venv set up and activated, and the server running. But you will see a warning in the terminal that no UI was found. If you go to the URL for the server, you won't get a UI.

docs/installation/manual.md

@@ -43,10 +43,10 @@ The following commands vary depending on the version of Invoke being installed a
 3. Create a virtual environment in that directory:
 
     ```sh
-    uv venv --relocatable --prompt invoke --python 3.12 --python-preference only-managed .venv
+    uv venv --relocatable --prompt invoke --python 3.11 --python-preference only-managed .venv
     ```
 
-    This command creates a portable virtual environment at `.venv` complete with a portable python 3.12. It doesn't matter if your system has no python installed, or has a different version - `uv` will handle everything.
+    This command creates a portable virtual environment at `.venv` complete with a portable python 3.11. It doesn't matter if your system has no python installed, or has a different version - `uv` will handle everything.
 
 4. Activate the virtual environment:
 
@@ -88,13 +88,13 @@ The following commands vary depending on the version of Invoke being installed a
 8. Install the `invokeai` package. Substitute the package specifier and version.
 
     ```sh
-    uv pip install <PACKAGE_SPECIFIER>==<VERSION> --python 3.12 --python-preference only-managed --force-reinstall
+    uv pip install <PACKAGE_SPECIFIER>==<VERSION> --python 3.11 --python-preference only-managed --force-reinstall
     ```
 
     If you determined you needed to use a `PyPI` index URL in the previous step, you'll need to add `--index=<INDEX_URL>` like this:
 
     ```sh
-    uv pip install <PACKAGE_SPECIFIER>==<VERSION> --python 3.12 --python-preference only-managed --index=<INDEX_URL> --force-reinstall
+    uv pip install <PACKAGE_SPECIFIER>==<VERSION> --python 3.11 --python-preference only-managed --index=<INDEX_URL> --force-reinstall
     ```
 
 9. Deactivate and reactivate your venv so that the invokeai-specific commands become available in the environment:

docs/installation/requirements.md

@@ -41,7 +41,7 @@ The requirements below are rough guidelines for best performance. GPUs with less
 
     You don't need to do this if you are installing with the [Invoke Launcher](./quick_start.md).
 
-Invoke requires python 3.10 through 3.12. If you don't already have one of these versions installed, we suggest installing 3.12, as it will be supported for longer.
+Invoke requires python 3.10 or 3.11. If you don't already have one of these versions installed, we suggest installing 3.11, as it will be supported for longer.
 
 Check that your system has an up-to-date Python installed by running `python3 --version` in the terminal (Linux, macOS) or cmd/powershell (Windows).
 
@@ -49,19 +49,19 @@ Check that your system has an up-to-date Python installed by running `python3 --
 
     === "Windows"
 
-        - Install python with [an official installer].
+        - Install python 3.11 with [an official installer].
         - The installer includes an option to add python to your PATH. Be sure to enable this. If you missed it, re-run the installer, choose to modify an existing installation, and tick that checkbox.
         - You may need to install [Microsoft Visual C++ Redistributable].
 
     === "macOS"
 
-        - Install python with [an official installer].
+        - Install python 3.11 with [an official installer].
         - If model installs fail with a certificate error, you may need to run this command (changing the python version to match what you have installed): `/Applications/Python\ 3.10/Install\ Certificates.command`
         - If you haven't already, you will need to install the XCode CLI Tools by running `xcode-select --install` in a terminal.
 
     === "Linux"
 
-        - Installing python varies depending on your system. We recommend [using `uv` to manage your python installation](https://docs.astral.sh/uv/concepts/python-versions/#installing-a-python-version).
+        - Installing python varies depending on your system. On Ubuntu, you can use the [deadsnakes PPA](https://launchpad.net/~deadsnakes/+archive/ubuntu/ppa).
         - You'll need to install `libglib2.0-0` and `libgl1-mesa-glx` for OpenCV to work. For example, on a Debian system: `sudo apt update && sudo apt install -y libglib2.0-0 libgl1-mesa-glx`
 
 ## Drivers

invokeai/app/api/dependencies.py

@@ -37,13 +37,7 @@ from invokeai.app.services.style_preset_records.style_preset_records_sqlite impo
 from invokeai.app.services.urls.urls_default import LocalUrlService
 from invokeai.app.services.workflow_records.workflow_records_sqlite import SqliteWorkflowRecordsStorage
 from invokeai.app.services.workflow_thumbnails.workflow_thumbnails_disk import WorkflowThumbnailFileStorageDisk
-from invokeai.backend.stable_diffusion.diffusion.conditioning_data import (
-    BasicConditioningInfo,
-    ConditioningFieldData,
-    FLUXConditioningInfo,
-    SD3ConditioningInfo,
-    SDXLConditioningInfo,
-)
+from invokeai.backend.stable_diffusion.diffusion.conditioning_data import ConditioningFieldData
 from invokeai.backend.util.logging import InvokeAILogger
 from invokeai.version.invokeai_version import __version__
 
@@ -107,25 +101,10 @@ class ApiDependencies:
         images = ImageService()
         invocation_cache = MemoryInvocationCache(max_cache_size=config.node_cache_size)
         tensors = ObjectSerializerForwardCache(
-            ObjectSerializerDisk[torch.Tensor](
-                output_folder / "tensors",
-                safe_globals=[torch.Tensor],
-                ephemeral=True,
-            ),
-            max_cache_size=0,
+            ObjectSerializerDisk[torch.Tensor](output_folder / "tensors", ephemeral=True)
         )
         conditioning = ObjectSerializerForwardCache(
-            ObjectSerializerDisk[ConditioningFieldData](
-                output_folder / "conditioning",
-                safe_globals=[
-                    ConditioningFieldData,
-                    BasicConditioningInfo,
-                    SDXLConditioningInfo,
-                    FLUXConditioningInfo,
-                    SD3ConditioningInfo,
-                ],
-                ephemeral=True,
-            ),
+            ObjectSerializerDisk[ConditioningFieldData](output_folder / "conditioning", ephemeral=True)
         )
         download_queue_service = DownloadQueueService(app_config=configuration, event_bus=events)
         model_images_service = ModelImageFileStorageDisk(model_images_folder / "model_images")

invokeai/app/invocations/baseinvocation.py

@@ -8,7 +8,6 @@ import sys
 import warnings
 from abc import ABC, abstractmethod
 from enum import Enum
-from functools import lru_cache
 from inspect import signature
 from typing import (
     TYPE_CHECKING,
@@ -28,6 +27,7 @@ import semver
 from pydantic import BaseModel, ConfigDict, Field, TypeAdapter, create_model
 from pydantic.fields import FieldInfo
 from pydantic_core import PydanticUndefined
+from typing_extensions import TypeAliasType
 
 from invokeai.app.invocations.fields import (
     FieldKind,
@@ -100,6 +100,37 @@ class BaseInvocationOutput(BaseModel):
     All invocation outputs must use the `@invocation_output` decorator to provide their unique type.
     """
 
+    _output_classes: ClassVar[set[BaseInvocationOutput]] = set()
+    _typeadapter: ClassVar[Optional[TypeAdapter[Any]]] = None
+    _typeadapter_needs_update: ClassVar[bool] = False
+
+    @classmethod
+    def register_output(cls, output: BaseInvocationOutput) -> None:
+        """Registers an invocation output."""
+        cls._output_classes.add(output)
+        cls._typeadapter_needs_update = True
+
+    @classmethod
+    def get_outputs(cls) -> Iterable[BaseInvocationOutput]:
+        """Gets all invocation outputs."""
+        return cls._output_classes
+
+    @classmethod
+    def get_typeadapter(cls) -> TypeAdapter[Any]:
+        """Gets a pydantc TypeAdapter for the union of all invocation output types."""
+        if not cls._typeadapter or cls._typeadapter_needs_update:
+            AnyInvocationOutput = TypeAliasType(
+                "AnyInvocationOutput", Annotated[Union[tuple(cls._output_classes)], Field(discriminator="type")]
+            )
+            cls._typeadapter = TypeAdapter(AnyInvocationOutput)
+            cls._typeadapter_needs_update = False
+        return cls._typeadapter
+
+    @classmethod
+    def get_output_types(cls) -> Iterable[str]:
+        """Gets all invocation output types."""
+        return (i.get_type() for i in BaseInvocationOutput.get_outputs())
+
     @staticmethod
     def json_schema_extra(schema: dict[str, Any], model_class: Type[BaseInvocationOutput]) -> None:
         """Adds various UI-facing attributes to the invocation output's OpenAPI schema."""
@@ -142,16 +173,76 @@ class BaseInvocation(ABC, BaseModel):
     All invocations must use the `@invocation` decorator to provide their unique type.
     """
 
+    _invocation_classes: ClassVar[set[BaseInvocation]] = set()
+    _typeadapter: ClassVar[Optional[TypeAdapter[Any]]] = None
+    _typeadapter_needs_update: ClassVar[bool] = False
+
     @classmethod
     def get_type(cls) -> str:
         """Gets the invocation's type, as provided by the `@invocation` decorator."""
         return cls.model_fields["type"].default
 
+    @classmethod
+    def register_invocation(cls, invocation: BaseInvocation) -> None:
+        """Registers an invocation."""
+        cls._invocation_classes.add(invocation)
+        cls._typeadapter_needs_update = True
+
+    @classmethod
+    def get_typeadapter(cls) -> TypeAdapter[Any]:
+        """Gets a pydantc TypeAdapter for the union of all invocation types."""
+        if not cls._typeadapter or cls._typeadapter_needs_update:
+            AnyInvocation = TypeAliasType(
+                "AnyInvocation", Annotated[Union[tuple(cls.get_invocations())], Field(discriminator="type")]
+            )
+            cls._typeadapter = TypeAdapter(AnyInvocation)
+            cls._typeadapter_needs_update = False
+        return cls._typeadapter
+
+    @classmethod
+    def invalidate_typeadapter(cls) -> None:
+        """Invalidates the typeadapter, forcing it to be rebuilt on next access. If the invocation allowlist or
+        denylist is changed, this should be called to ensure the typeadapter is updated and validation respects
+        the updated allowlist and denylist."""
+        cls._typeadapter_needs_update = True
+
+    @classmethod
+    def get_invocations(cls) -> Iterable[BaseInvocation]:
+        """Gets all invocations, respecting the allowlist and denylist."""
+        app_config = get_config()
+        allowed_invocations: set[BaseInvocation] = set()
+        for sc in cls._invocation_classes:
+            invocation_type = sc.get_type()
+            is_in_allowlist = (
+                invocation_type in app_config.allow_nodes if isinstance(app_config.allow_nodes, list) else True
+            )
+            is_in_denylist = (
+                invocation_type in app_config.deny_nodes if isinstance(app_config.deny_nodes, list) else False
+            )
+            if is_in_allowlist and not is_in_denylist:
+                allowed_invocations.add(sc)
+        return allowed_invocations
+
+    @classmethod
+    def get_invocations_map(cls) -> dict[str, BaseInvocation]:
+        """Gets a map of all invocation types to their invocation classes."""
+        return {i.get_type(): i for i in BaseInvocation.get_invocations()}
+
+    @classmethod
+    def get_invocation_types(cls) -> Iterable[str]:
+        """Gets all invocation types."""
+        return (i.get_type() for i in BaseInvocation.get_invocations())
+
     @classmethod
     def get_output_annotation(cls) -> BaseInvocationOutput:
         """Gets the invocation's output annotation (i.e. the return annotation of its `invoke()` method)."""
         return signature(cls.invoke).return_annotation
 
+    @classmethod
+    def get_invocation_for_type(cls, invocation_type: str) -> BaseInvocation | None:
+        """Gets the invocation class for a given invocation type."""
+        return cls.get_invocations_map().get(invocation_type)
+
     @staticmethod
     def json_schema_extra(schema: dict[str, Any], model_class: Type[BaseInvocation]) -> None:
         """Adds various UI-facing attributes to the invocation's OpenAPI schema."""
@@ -249,105 +340,6 @@ class BaseInvocation(ABC, BaseModel):
 TBaseInvocation = TypeVar("TBaseInvocation", bound=BaseInvocation)
 
 
-class InvocationRegistry:
-    _invocation_classes: ClassVar[set[type[BaseInvocation]]] = set()
-    _output_classes: ClassVar[set[type[BaseInvocationOutput]]] = set()
-
-    @classmethod
-    def register_invocation(cls, invocation: type[BaseInvocation]) -> None:
-        """Registers an invocation."""
-        cls._invocation_classes.add(invocation)
-        cls.invalidate_invocation_typeadapter()
-
-    @classmethod
-    @lru_cache(maxsize=1)
-    def get_invocation_typeadapter(cls) -> TypeAdapter[Any]:
-        """Gets a pydantic TypeAdapter for the union of all invocation types.
-
-        This is used to parse serialized invocations into the correct invocation class.
-
-        This method is cached to avoid rebuilding the TypeAdapter on every access. If the invocation allowlist or
-        denylist is changed, the cache should be cleared to ensure the TypeAdapter is updated and validation respects
-        the updated allowlist and denylist.
-
-        @see https://docs.pydantic.dev/latest/concepts/type_adapter/
-        """
-        return TypeAdapter(Annotated[Union[tuple(cls.get_invocation_classes())], Field(discriminator="type")])
-
-    @classmethod
-    def invalidate_invocation_typeadapter(cls) -> None:
-        """Invalidates the cached invocation type adapter."""
-        cls.get_invocation_typeadapter.cache_clear()
-
-    @classmethod
-    def get_invocation_classes(cls) -> Iterable[type[BaseInvocation]]:
-        """Gets all invocations, respecting the allowlist and denylist."""
-        app_config = get_config()
-        allowed_invocations: set[type[BaseInvocation]] = set()
-        for sc in cls._invocation_classes:
-            invocation_type = sc.get_type()
-            is_in_allowlist = (
-                invocation_type in app_config.allow_nodes if isinstance(app_config.allow_nodes, list) else True
-            )
-            is_in_denylist = (
-                invocation_type in app_config.deny_nodes if isinstance(app_config.deny_nodes, list) else False
-            )
-            if is_in_allowlist and not is_in_denylist:
-                allowed_invocations.add(sc)
-        return allowed_invocations
-
-    @classmethod
-    def get_invocations_map(cls) -> dict[str, type[BaseInvocation]]:
-        """Gets a map of all invocation types to their invocation classes."""
-        return {i.get_type(): i for i in cls.get_invocation_classes()}
-
-    @classmethod
-    def get_invocation_types(cls) -> Iterable[str]:
-        """Gets all invocation types."""
-        return (i.get_type() for i in cls.get_invocation_classes())
-
-    @classmethod
-    def get_invocation_for_type(cls, invocation_type: str) -> type[BaseInvocation] | None:
-        """Gets the invocation class for a given invocation type."""
-        return cls.get_invocations_map().get(invocation_type)
-
-    @classmethod
-    def register_output(cls, output: "type[TBaseInvocationOutput]") -> None:
-        """Registers an invocation output."""
-        cls._output_classes.add(output)
-        cls.invalidate_output_typeadapter()
-
-    @classmethod
-    def get_output_classes(cls) -> Iterable[type[BaseInvocationOutput]]:
-        """Gets all invocation outputs."""
-        return cls._output_classes
-
-    @classmethod
-    @lru_cache(maxsize=1)
-    def get_output_typeadapter(cls) -> TypeAdapter[Any]:
-        """Gets a pydantic TypeAdapter for the union of all invocation output types.
-
-        This is used to parse serialized invocation outputs into the correct invocation output class.
-
-        This method is cached to avoid rebuilding the TypeAdapter on every access. If the invocation allowlist or
-        denylist is changed, the cache should be cleared to ensure the TypeAdapter is updated and validation respects
-        the updated allowlist and denylist.
-
-        @see https://docs.pydantic.dev/latest/concepts/type_adapter/
-        """
-        return TypeAdapter(Annotated[Union[tuple(cls._output_classes)], Field(discriminator="type")])
-
-    @classmethod
-    def invalidate_output_typeadapter(cls) -> None:
-        """Invalidates the cached invocation output type adapter."""
-        cls.get_output_typeadapter.cache_clear()
-
-    @classmethod
-    def get_output_types(cls) -> Iterable[str]:
-        """Gets all invocation output types."""
-        return (i.get_type() for i in cls.get_output_classes())
-
-
 RESERVED_NODE_ATTRIBUTE_FIELD_NAMES = {
     "id",
     "is_intermediate",
@@ -461,8 +453,8 @@ def invocation(
         node_pack = cls.__module__.split(".")[0]
 
         # Handle the case where an existing node is being clobbered by the one we are registering
-        if invocation_type in InvocationRegistry.get_invocation_types():
-            clobbered_invocation = InvocationRegistry.get_invocation_for_type(invocation_type)
+        if invocation_type in BaseInvocation.get_invocation_types():
+            clobbered_invocation = BaseInvocation.get_invocation_for_type(invocation_type)
             # This should always be true - we just checked if the invocation type was in the set
             assert clobbered_invocation is not None
 
@@ -547,7 +539,8 @@ def invocation(
         )
         cls.__doc__ = docstring
 
-        InvocationRegistry.register_invocation(cls)
+        # TODO: how to type this correctly? it's typed as ModelMetaclass, a private class in pydantic
+        BaseInvocation.register_invocation(cls)  # type: ignore
 
         return cls
 
@@ -572,7 +565,7 @@ def invocation_output(
         if re.compile(r"^\S+$").match(output_type) is None:
             raise ValueError(f'"output_type" must consist of non-whitespace characters, got "{output_type}"')
 
-        if output_type in InvocationRegistry.get_output_types():
+        if output_type in BaseInvocationOutput.get_output_types():
             raise ValueError(f'Invocation type "{output_type}" already exists')
 
         validate_fields(cls.model_fields, output_type)
@@ -593,7 +586,7 @@ def invocation_output(
         )
         cls.__doc__ = docstring
 
-        InvocationRegistry.register_output(cls)
+        BaseInvocationOutput.register_output(cls)  # type: ignore # TODO: how to type this correctly?
 
         return cls
 

invokeai/app/invocations/image.py

@@ -1089,7 +1089,7 @@ class CanvasV2MaskAndCropInvocation(BaseInvocation, WithMetadata, WithBoard):
 
 
 @invocation(
-    "expand_mask_with_fade", title="Expand Mask with Fade", tags=["image", "mask"], category="image", version="1.0.1"
+    "expand_mask_with_fade", title="Expand Mask with Fade", tags=["image", "mask"], category="image", version="1.0.0"
 )
 class ExpandMaskWithFadeInvocation(BaseInvocation, WithMetadata, WithBoard):
     """Expands a mask with a fade effect. The mask uses black to indicate areas to keep from the generated image and white for areas to discard.
@@ -1147,21 +1147,8 @@ class ExpandMaskWithFadeInvocation(BaseInvocation, WithMetadata, WithBoard):
         coeffs = numpy.polyfit(x_control, y_control, 3)
         poly = numpy.poly1d(coeffs)
 
-        # Evaluate the polynomial
-        feather = poly(d_norm)
-
-        # The polynomial fit isn't perfect. Points beyond the fade distance are likely to be slightly less than 1.0,
-        # even though the control points indicate that they should be exactly 1.0. This is due to the nature of the
-        # polynomial fit, which is a best approximation of the control points but not an exact match.
-
-        # When this occurs, the area outside the mask and fade-out will not be 100% transparent. For example, it may
-        # have an alpha value of 1 instead of 0. So we must force pixels at or beyond the fade distance to exactly 1.0.
-
-        # Force pixels at or beyond the fade distance to exactly 1.0
-        feather = numpy.where(d_norm >= 1.0, 1.0, feather)
-
-        # Clip any other values to ensure they're in the valid range [0,1]
-        feather = numpy.clip(feather, 0, 1)
+        # Evaluate and clip the smooth mapping
+        feather = numpy.clip(poly(d_norm), 0, 1)
 
         # Build final image.
         np_result = numpy.where(black_mask == 1, 0, (feather * 255).astype(numpy.uint8))

invokeai/app/services/object_serializer/object_serializer_disk.py

@@ -21,16 +21,10 @@ class ObjectSerializerDisk(ObjectSerializerBase[T]):
     """Disk-backed storage for arbitrary python objects. Serialization is handled by `torch.save` and `torch.load`.
 
     :param output_dir: The folder where the serialized objects will be stored
-    :param safe_globals: A list of types to be added to the safe globals for torch serialization
     :param ephemeral: If True, objects will be stored in a temporary directory inside the given output_dir and cleaned up on exit
     """
 
-    def __init__(
-        self,
-        output_dir: Path,
-        safe_globals: list[type],
-        ephemeral: bool = False,
-    ) -> None:
+    def __init__(self, output_dir: Path, ephemeral: bool = False):
         super().__init__()
         self._ephemeral = ephemeral
         self._base_output_dir = output_dir
@@ -48,8 +42,6 @@ class ObjectSerializerDisk(ObjectSerializerBase[T]):
         self._output_dir = Path(self._tempdir.name) if self._tempdir else self._base_output_dir
         self.__obj_class_name: Optional[str] = None
 
-        torch.serialization.add_safe_globals(safe_globals) if safe_globals else None
-
     def load(self, name: str) -> T:
         file_path = self._get_path(name)
         try:

invokeai/app/services/shared/graph.py

@@ -21,7 +21,6 @@ from invokeai.app.invocations import *  # noqa: F401 F403
 from invokeai.app.invocations.baseinvocation import (
     BaseInvocation,
     BaseInvocationOutput,
-    InvocationRegistry,
     invocation,
     invocation_output,
 )
@@ -284,7 +283,7 @@ class AnyInvocation(BaseInvocation):
     @classmethod
     def __get_pydantic_core_schema__(cls, source_type: Any, handler: GetCoreSchemaHandler) -> core_schema.CoreSchema:
         def validate_invocation(v: Any) -> "AnyInvocation":
-            return InvocationRegistry.get_invocation_typeadapter().validate_python(v)
+            return BaseInvocation.get_typeadapter().validate_python(v)
 
         return core_schema.no_info_plain_validator_function(validate_invocation)
 
@@ -295,7 +294,7 @@ class AnyInvocation(BaseInvocation):
         # Nodes are too powerful, we have to make our own OpenAPI schema manually
         # No but really, because the schema is dynamic depending on loaded nodes, we need to generate it manually
         oneOf: list[dict[str, str]] = []
-        names = [i.__name__ for i in InvocationRegistry.get_invocation_classes()]
+        names = [i.__name__ for i in BaseInvocation.get_invocations()]
         for name in sorted(names):
             oneOf.append({"$ref": f"#/components/schemas/{name}"})
         return {"oneOf": oneOf}
@@ -305,7 +304,7 @@ class AnyInvocationOutput(BaseInvocationOutput):
     @classmethod
     def __get_pydantic_core_schema__(cls, source_type: Any, handler: GetCoreSchemaHandler):
         def validate_invocation_output(v: Any) -> "AnyInvocationOutput":
-            return InvocationRegistry.get_output_typeadapter().validate_python(v)
+            return BaseInvocationOutput.get_typeadapter().validate_python(v)
 
         return core_schema.no_info_plain_validator_function(validate_invocation_output)
 
@@ -317,7 +316,7 @@ class AnyInvocationOutput(BaseInvocationOutput):
         # No but really, because the schema is dynamic depending on loaded nodes, we need to generate it manually
 
         oneOf: list[dict[str, str]] = []
-        names = [i.__name__ for i in InvocationRegistry.get_output_classes()]
+        names = [i.__name__ for i in BaseInvocationOutput.get_outputs()]
         for name in sorted(names):
             oneOf.append({"$ref": f"#/components/schemas/{name}"})
         return {"oneOf": oneOf}

invokeai/app/util/custom_openapi.py

@@ -4,10 +4,7 @@ from fastapi import FastAPI
 from fastapi.openapi.utils import get_openapi
 from pydantic.json_schema import models_json_schema
 
-from invokeai.app.invocations.baseinvocation import (
-    InvocationRegistry,
-    UIConfigBase,
-)
+from invokeai.app.invocations.baseinvocation import BaseInvocation, BaseInvocationOutput, UIConfigBase
 from invokeai.app.invocations.fields import InputFieldJSONSchemaExtra, OutputFieldJSONSchemaExtra
 from invokeai.app.invocations.model import ModelIdentifierField
 from invokeai.app.services.events.events_common import EventBase
@@ -59,14 +56,14 @@ def get_openapi_func(
         invocation_output_map_required: list[str] = []
 
         # We need to manually add all outputs to the schema - pydantic doesn't add them because they aren't used directly.
-        for output in InvocationRegistry.get_output_classes():
+        for output in BaseInvocationOutput.get_outputs():
             json_schema = output.model_json_schema(mode="serialization", ref_template="#/components/schemas/{model}")
             move_defs_to_top_level(openapi_schema, json_schema)
             openapi_schema["components"]["schemas"][output.__name__] = json_schema
 
         # Technically, invocations are added to the schema by pydantic, but we still need to manually set their output
         # property, so we'll just do it all manually.
-        for invocation in InvocationRegistry.get_invocation_classes():
+        for invocation in BaseInvocation.get_invocations():
             json_schema = invocation.model_json_schema(
                 mode="serialization", ref_template="#/components/schemas/{model}"
             )

invokeai/backend/flux/flux_state_dict_utils.py

@@ -1,23 +0,0 @@
-from typing import TYPE_CHECKING
-
-if TYPE_CHECKING:
-    from invokeai.backend.model_manager.legacy_probe import CkptType
-
-
-def get_flux_in_channels_from_state_dict(state_dict: "CkptType") -> int | None:
-    """Gets the in channels from the state dict."""
-
-    # "Standard" FLUX models use "img_in.weight", but some community fine tunes use
-    # "model.diffusion_model.img_in.weight". Known models that use the latter key:
-    # - https://civitai.com/models/885098?modelVersionId=990775
-    # - https://civitai.com/models/1018060?modelVersionId=1596255
-    # - https://civitai.com/models/978314/ultrareal-fine-tune?modelVersionId=1413133
-
-    keys = {"img_in.weight", "model.diffusion_model.img_in.weight"}
-
-    for key in keys:
-        val = state_dict.get(key)
-        if val is not None:
-            return val.shape[1]
-
-    return None

invokeai/backend/model_manager/config.py

@@ -30,18 +30,19 @@ from inspect import isabstract
 from pathlib import Path
 from typing import ClassVar, Literal, Optional, TypeAlias, Union
 
+import safetensors.torch
+import torch
+from picklescan.scanner import scan_file_path
 from pydantic import BaseModel, ConfigDict, Discriminator, Field, Tag, TypeAdapter
 from typing_extensions import Annotated, Any, Dict
 
 from invokeai.app.util.misc import uuid_string
 from invokeai.backend.model_hash.hash_validator import validate_hash
-from invokeai.backend.model_hash.model_hash import HASHING_ALGORITHMS
-from invokeai.backend.model_manager.model_on_disk import ModelOnDisk
+from invokeai.backend.model_hash.model_hash import HASHING_ALGORITHMS, ModelHash
 from invokeai.backend.model_manager.taxonomy import (
     AnyVariant,
     BaseModelType,
     ClipVariantType,
-    FluxLoRAFormat,
     ModelFormat,
     ModelRepoVariant,
     ModelSourceType,
@@ -50,8 +51,9 @@ from invokeai.backend.model_manager.taxonomy import (
     SchedulerPredictionType,
     SubModelType,
 )
-from invokeai.backend.model_manager.util.model_util import lora_token_vector_length
+from invokeai.backend.quantization.gguf.loaders import gguf_sd_loader
 from invokeai.backend.stable_diffusion.schedulers.schedulers import SCHEDULER_NAME_VALUES
+from invokeai.backend.util.silence_warnings import SilenceWarnings
 
 logger = logging.getLogger(__name__)
 
@@ -65,6 +67,11 @@ class InvalidModelConfigException(Exception):
 DEFAULTS_PRECISION = Literal["fp16", "fp32"]
 
 
+class FSLayout(Enum):
+    FILE = "file"
+    DIRECTORY = "directory"
+
+
 class SubmodelDefinition(BaseModel):
     path_or_prefix: str
     model_type: ModelType
@@ -95,6 +102,87 @@ class ControlAdapterDefaultSettings(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
 
+class ModelOnDisk:
+    """A utility class representing a model stored on disk."""
+
+    def __init__(self, path: Path, hash_algo: HASHING_ALGORITHMS = "blake3_single"):
+        self.path = path
+        # TODO: Revisit checkpoint vs diffusers terminology
+        self.layout = FSLayout.DIRECTORY if path.is_dir() else FSLayout.FILE
+        if self.path.suffix in {".safetensors", ".bin", ".pt", ".ckpt"}:
+            self.name = path.stem
+        else:
+            self.name = path.name
+        self.hash_algo = hash_algo
+        self._state_dict_cache = {}
+
+    def hash(self) -> str:
+        return ModelHash(algorithm=self.hash_algo).hash(self.path)
+
+    def size(self) -> int:
+        if self.layout == FSLayout.FILE:
+            return self.path.stat().st_size
+        return sum(file.stat().st_size for file in self.path.rglob("*"))
+
+    def component_paths(self) -> set[Path]:
+        if self.layout == FSLayout.FILE:
+            return {self.path}
+        extensions = {".safetensors", ".pt", ".pth", ".ckpt", ".bin", ".gguf"}
+        return {f for f in self.path.rglob("*") if f.suffix in extensions}
+
+    def repo_variant(self) -> Optional[ModelRepoVariant]:
+        if self.layout == FSLayout.FILE:
+            return None
+
+        weight_files = list(self.path.glob("**/*.safetensors"))
+        weight_files.extend(list(self.path.glob("**/*.bin")))
+        for x in weight_files:
+            if ".fp16" in x.suffixes:
+                return ModelRepoVariant.FP16
+            if "openvino_model" in x.name:
+                return ModelRepoVariant.OpenVINO
+            if "flax_model" in x.name:
+                return ModelRepoVariant.Flax
+            if x.suffix == ".onnx":
+                return ModelRepoVariant.ONNX
+        return ModelRepoVariant.Default
+
+    def load_state_dict(self, path: Optional[Path] = None) -> Dict[str | int, Any]:
+        if path in self._state_dict_cache:
+            return self._state_dict_cache[path]
+
+        if not path:
+            components = list(self.component_paths())
+            match components:
+                case []:
+                    raise ValueError("No weight files found for this model")
+                case [p]:
+                    path = p
+                case ps if len(ps) >= 2:
+                    raise ValueError(
+                        f"Multiple weight files found for this model: {ps}. "
+                        f"Please specify the intended file using the 'path' argument"
+                    )
+
+        with SilenceWarnings():
+            if path.suffix.endswith((".ckpt", ".pt", ".pth", ".bin")):
+                scan_result = scan_file_path(path)
+                if scan_result.infected_files != 0 or scan_result.scan_err:
+                    raise RuntimeError(f"The model {path.stem} is potentially infected by malware. Aborting import.")
+                checkpoint = torch.load(path, map_location="cpu")
+                assert isinstance(checkpoint, dict)
+            elif path.suffix.endswith(".gguf"):
+                checkpoint = gguf_sd_loader(path, compute_dtype=torch.float32)
+            elif path.suffix.endswith(".safetensors"):
+                checkpoint = safetensors.torch.load_file(path)
+            else:
+                raise ValueError(f"Unrecognized model extension: {path.suffix}")
+
+        state_dict = checkpoint.get("state_dict", checkpoint)
+        self._state_dict_cache[path] = state_dict
+        return state_dict
+
+
 class MatchSpeed(int, Enum):
     """Represents the estimated runtime speed of a config's 'matches' method."""
 
@@ -169,7 +257,7 @@ class ModelConfigBase(ABC, BaseModel):
         Created to deprecate ModelProbe.probe
         """
         candidates = ModelConfigBase._USING_CLASSIFY_API
-        sorted_by_match_speed = sorted(candidates, key=lambda cls: (cls._MATCH_SPEED, cls.__name__))
+        sorted_by_match_speed = sorted(candidates, key=lambda cls: cls._MATCH_SPEED)
         mod = ModelOnDisk(model_path, hash_algo)
 
         for config_cls in sorted_by_match_speed:
@@ -220,9 +308,6 @@ class ModelConfigBase(ABC, BaseModel):
         if "source_type" in overrides:
             overrides["source_type"] = ModelSourceType(overrides["source_type"])
 
-        if "variant" in overrides:
-            overrides["variant"] = ModelVariantType(overrides["variant"])
-
     @classmethod
     def from_model_on_disk(cls, mod: ModelOnDisk, **overrides):
         """Creates an instance of this config or raises InvalidModelConfigException."""
@@ -282,38 +367,6 @@ class LoRAConfigBase(ABC, BaseModel):
     type: Literal[ModelType.LoRA] = ModelType.LoRA
     trigger_phrases: Optional[set[str]] = Field(description="Set of trigger phrases for this model", default=None)
 
-    @classmethod
-    def flux_lora_format(cls, mod: ModelOnDisk):
-        key = "FLUX_LORA_FORMAT"
-        if key in mod.cache:
-            return mod.cache[key]
-
-        from invokeai.backend.patches.lora_conversions.formats import flux_format_from_state_dict
-
-        sd = mod.load_state_dict(mod.path)
-        value = flux_format_from_state_dict(sd)
-        mod.cache[key] = value
-        return value
-
-    @classmethod
-    def base_model(cls, mod: ModelOnDisk) -> BaseModelType:
-        if cls.flux_lora_format(mod):
-            return BaseModelType.Flux
-
-        state_dict = mod.load_state_dict()
-        # If we've gotten here, we assume that the model is a Stable Diffusion model
-        token_vector_length = lora_token_vector_length(state_dict)
-        if token_vector_length == 768:
-            return BaseModelType.StableDiffusion1
-        elif token_vector_length == 1024:
-            return BaseModelType.StableDiffusion2
-        elif token_vector_length == 1280:
-            return BaseModelType.StableDiffusionXL  # recognizes format at https://civitai.com/models/224641
-        elif token_vector_length == 2048:
-            return BaseModelType.StableDiffusionXL
-        else:
-            raise InvalidModelConfigException("Unknown LoRA type")
-
 
 class T5EncoderConfigBase(ABC, BaseModel):
     """Base class for diffusers-style models."""
@@ -329,40 +382,11 @@ class T5EncoderBnbQuantizedLlmInt8bConfig(T5EncoderConfigBase, LegacyProbeMixin,
     format: Literal[ModelFormat.BnbQuantizedLlmInt8b] = ModelFormat.BnbQuantizedLlmInt8b
 
 
-class LoRALyCORISConfig(LoRAConfigBase, ModelConfigBase):
+class LoRALyCORISConfig(LoRAConfigBase, LegacyProbeMixin, ModelConfigBase):
     """Model config for LoRA/Lycoris models."""
 
     format: Literal[ModelFormat.LyCORIS] = ModelFormat.LyCORIS
 
-    @classmethod
-    def matches(cls, mod: ModelOnDisk) -> bool:
-        if mod.path.is_dir():
-            return False
-
-        # Avoid false positive match against ControlLoRA and Diffusers
-        if cls.flux_lora_format(mod) in [FluxLoRAFormat.Control, FluxLoRAFormat.Diffusers]:
-            return False
-
-        state_dict = mod.load_state_dict()
-        for key in state_dict.keys():
-            if type(key) is int:
-                continue
-
-            if key.startswith(("lora_te_", "lora_unet_", "lora_te1_", "lora_te2_", "lora_transformer_")):
-                return True
-            # "lora_A.weight" and "lora_B.weight" are associated with models in PEFT format. We don't support all PEFT
-            # LoRA models, but as of the time of writing, we support Diffusers FLUX PEFT LoRA models.
-            if key.endswith(("to_k_lora.up.weight", "to_q_lora.down.weight", "lora_A.weight", "lora_B.weight")):
-                return True
-
-        return False
-
-    @classmethod
-    def parse(cls, mod: ModelOnDisk) -> dict[str, Any]:
-        return {
-            "base": cls.base_model(mod),
-        }
-
 
 class ControlAdapterConfigBase(ABC, BaseModel):
     default_settings: Optional[ControlAdapterDefaultSettings] = Field(
@@ -386,26 +410,11 @@ class ControlLoRADiffusersConfig(ControlAdapterConfigBase, LegacyProbeMixin, Mod
     format: Literal[ModelFormat.Diffusers] = ModelFormat.Diffusers
 
 
-class LoRADiffusersConfig(LoRAConfigBase, ModelConfigBase):
+class LoRADiffusersConfig(LoRAConfigBase, LegacyProbeMixin, ModelConfigBase):
     """Model config for LoRA/Diffusers models."""
 
     format: Literal[ModelFormat.Diffusers] = ModelFormat.Diffusers
 
-    @classmethod
-    def matches(cls, mod: ModelOnDisk) -> bool:
-        if mod.path.is_file():
-            return cls.flux_lora_format(mod) == FluxLoRAFormat.Diffusers
-
-        suffixes = ["bin", "safetensors"]
-        weight_files = [mod.path / f"pytorch_lora_weights.{sfx}" for sfx in suffixes]
-        return any(wf.exists() for wf in weight_files)
-
-    @classmethod
-    def parse(cls, mod: ModelOnDisk) -> dict[str, Any]:
-        return {
-            "base": cls.base_model(mod),
-        }
-
 
 class VAECheckpointConfig(CheckpointConfigBase, LegacyProbeMixin, ModelConfigBase):
     """Model config for standalone VAE models."""
@@ -577,7 +586,7 @@ class LlavaOnevisionConfig(DiffusersConfigBase, ModelConfigBase):
 
     @classmethod
     def matches(cls, mod: ModelOnDisk) -> bool:
-        if mod.path.is_file():
+        if mod.layout == FSLayout.FILE:
             return False
 
         config_path = mod.path / "config.json"

invokeai/backend/model_manager/legacy_probe.py

@@ -14,7 +14,6 @@ from invokeai.backend.flux.controlnet.state_dict_utils import (
     is_state_dict_instantx_controlnet,
     is_state_dict_xlabs_controlnet,
 )
-from invokeai.backend.flux.flux_state_dict_utils import get_flux_in_channels_from_state_dict
 from invokeai.backend.flux.ip_adapter.state_dict_utils import is_state_dict_xlabs_ip_adapter
 from invokeai.backend.flux.redux.flux_redux_state_dict_utils import is_state_dict_likely_flux_redux
 from invokeai.backend.model_hash.model_hash import HASHING_ALGORITHMS, ModelHash
@@ -565,14 +564,7 @@ class CheckpointProbeBase(ProbeBase):
         state_dict = self.checkpoint.get("state_dict") or self.checkpoint
 
         if base_type == BaseModelType.Flux:
-            in_channels = get_flux_in_channels_from_state_dict(state_dict)
-
-            if in_channels is None:
-                # If we cannot find the in_channels, we assume that this is a normal variant. Log a warning.
-                logger.warning(
-                    f"{self.model_path} does not have img_in.weight or model.diffusion_model.img_in.weight key. Assuming normal variant."
-                )
-                return ModelVariantType.Normal
+            in_channels = state_dict["img_in.weight"].shape[1]
 
             # FLUX Model variant types are distinguished by input channels:
             # - Unquantized Dev and Schnell have in_channels=64

invokeai/backend/model_manager/model_on_disk.py

@@ -1,96 +0,0 @@
-from pathlib import Path
-from typing import Any, Optional, TypeAlias
-
-import safetensors.torch
-import torch
-from picklescan.scanner import scan_file_path
-
-from invokeai.backend.model_hash.model_hash import HASHING_ALGORITHMS, ModelHash
-from invokeai.backend.model_manager.taxonomy import ModelRepoVariant
-from invokeai.backend.quantization.gguf.loaders import gguf_sd_loader
-from invokeai.backend.util.silence_warnings import SilenceWarnings
-
-StateDict: TypeAlias = dict[str | int, Any]  # When are the keys int?
-
-
-class ModelOnDisk:
-    """A utility class representing a model stored on disk."""
-
-    def __init__(self, path: Path, hash_algo: HASHING_ALGORITHMS = "blake3_single"):
-        self.path = path
-        if self.path.suffix in {".safetensors", ".bin", ".pt", ".ckpt"}:
-            self.name = path.stem
-        else:
-            self.name = path.name
-        self.hash_algo = hash_algo
-        # Having a cache helps users of ModelOnDisk (i.e. configs) to save state
-        # This prevents redundant computations during matching and parsing
-        self.cache = {"_CACHED_STATE_DICTS": {}}
-
-    def hash(self) -> str:
-        return ModelHash(algorithm=self.hash_algo).hash(self.path)
-
-    def size(self) -> int:
-        if self.path.is_file():
-            return self.path.stat().st_size
-        return sum(file.stat().st_size for file in self.path.rglob("*"))
-
-    def component_paths(self) -> set[Path]:
-        if self.path.is_file():
-            return {self.path}
-        extensions = {".safetensors", ".pt", ".pth", ".ckpt", ".bin", ".gguf"}
-        return {f for f in self.path.rglob("*") if f.suffix in extensions}
-
-    def repo_variant(self) -> Optional[ModelRepoVariant]:
-        if self.path.is_file():
-            return None
-
-        weight_files = list(self.path.glob("**/*.safetensors"))
-        weight_files.extend(list(self.path.glob("**/*.bin")))
-        for x in weight_files:
-            if ".fp16" in x.suffixes:
-                return ModelRepoVariant.FP16
-            if "openvino_model" in x.name:
-                return ModelRepoVariant.OpenVINO
-            if "flax_model" in x.name:
-                return ModelRepoVariant.Flax
-            if x.suffix == ".onnx":
-                return ModelRepoVariant.ONNX
-        return ModelRepoVariant.Default
-
-    def load_state_dict(self, path: Optional[Path] = None) -> StateDict:
-        sd_cache = self.cache["_CACHED_STATE_DICTS"]
-
-        if path in sd_cache:
-            return sd_cache[path]
-
-        if not path:
-            components = list(self.component_paths())
-            match components:
-                case []:
-                    raise ValueError("No weight files found for this model")
-                case [p]:
-                    path = p
-                case ps if len(ps) >= 2:
-                    raise ValueError(
-                        f"Multiple weight files found for this model: {ps}. "
-                        f"Please specify the intended file using the 'path' argument"
-                    )
-
-        with SilenceWarnings():
-            if path.suffix.endswith((".ckpt", ".pt", ".pth", ".bin")):
-                scan_result = scan_file_path(path)
-                if scan_result.infected_files != 0 or scan_result.scan_err:
-                    raise RuntimeError(f"The model {path.stem} is potentially infected by malware. Aborting import.")
-                checkpoint = torch.load(path, map_location="cpu")
-                assert isinstance(checkpoint, dict)
-            elif path.suffix.endswith(".gguf"):
-                checkpoint = gguf_sd_loader(path, compute_dtype=torch.float32)
-            elif path.suffix.endswith(".safetensors"):
-                checkpoint = safetensors.torch.load_file(path)
-            else:
-                raise ValueError(f"Unrecognized model extension: {path.suffix}")
-
-        state_dict = checkpoint.get("state_dict", checkpoint)
-        sd_cache[path] = state_dict
-        return state_dict

invokeai/backend/model_manager/taxonomy.py

@@ -126,13 +126,4 @@ class ModelSourceType(str, Enum):
     HFRepoID = "hf_repo_id"
 
 
-class FluxLoRAFormat(str, Enum):
-    """Flux LoRA formats."""
-
-    Diffusers = "flux.diffusers"
-    Kohya = "flux.kohya"
-    OneTrainer = "flux.onetrainer"
-    Control = "flux.control"
-
-
 AnyVariant: TypeAlias = Union[ModelVariantType, ClipVariantType, None]

invokeai/backend/patches/lora_conversions/formats.py

@@ -1,24 +0,0 @@
-from invokeai.backend.model_manager.taxonomy import FluxLoRAFormat
-from invokeai.backend.patches.lora_conversions.flux_control_lora_utils import is_state_dict_likely_flux_control
-from invokeai.backend.patches.lora_conversions.flux_diffusers_lora_conversion_utils import (
-    is_state_dict_likely_in_flux_diffusers_format,
-)
-from invokeai.backend.patches.lora_conversions.flux_kohya_lora_conversion_utils import (
-    is_state_dict_likely_in_flux_kohya_format,
-)
-from invokeai.backend.patches.lora_conversions.flux_onetrainer_lora_conversion_utils import (
-    is_state_dict_likely_in_flux_onetrainer_format,
-)
-
-
-def flux_format_from_state_dict(state_dict):
-    if is_state_dict_likely_in_flux_kohya_format(state_dict):
-        return FluxLoRAFormat.Kohya
-    elif is_state_dict_likely_in_flux_onetrainer_format(state_dict):
-        return FluxLoRAFormat.OneTrainer
-    elif is_state_dict_likely_in_flux_diffusers_format(state_dict):
-        return FluxLoRAFormat.Diffusers
-    elif is_state_dict_likely_flux_control(state_dict):
-        return FluxLoRAFormat.Control
-    else:
-        return None

invokeai/backend/stable_diffusion/diffusion/conditioning_data.py

@@ -69,9 +69,6 @@ class SD3ConditioningInfo:
 
 @dataclass
 class ConditioningFieldData:
-    # If you change this class, adding more types, you _must_ update the instantiation of ObjectSerializerDisk in
-    # invokeai/app/api/dependencies.py, adding the types to the list of safe globals. If you do not, torch will be
-    # unable to deserialize the object and will raise an error.
     conditionings: (
         List[BasicConditioningInfo]
         | List[SDXLConditioningInfo]

invokeai/frontend/web/package.json

@@ -62,7 +62,7 @@
     "@nanostores/react": "^0.7.3",
     "@reduxjs/toolkit": "2.6.1",
     "@roarr/browser-log-writer": "^1.3.0",
-    "@xyflow/react": "^12.5.3",
+    "@xyflow/react": "^12.4.2",
     "async-mutex": "^0.5.0",
     "chakra-react-select": "^4.9.2",
     "cmdk": "^1.0.0",
@@ -162,6 +162,5 @@
   },
   "engines": {
     "pnpm": "8"
-  },
-  "packageManager": "pnpm@8.15.9+sha512.499434c9d8fdd1a2794ebf4552b3b25c0a633abcee5bb15e7b5de90f32f47b513aca98cd5cfd001c31f0db454bc3804edccd578501e4ca293a6816166bbd9f81"
+  }
 }

invokeai/frontend/web/pnpm-lock.yaml

@@ -36,8 +36,8 @@ dependencies:
     specifier: ^1.3.0
     version: 1.3.0
   '@xyflow/react':
-    specifier: ^12.5.3
-    version: 12.5.3(@types/react@18.3.11)(react-dom@18.3.1)(react@18.3.1)
+    specifier: ^12.4.2
+    version: 12.4.2(@types/react@18.3.11)(react-dom@18.3.1)(react@18.3.1)
   async-mutex:
     specifier: ^0.5.0
     version: 0.5.0
@@ -3323,7 +3323,7 @@ packages:
   /@types/d3-drag@3.0.7:
     resolution: {integrity: sha512-HE3jVKlzU9AaMazNufooRJ5ZpWmLIoc90A37WU2JMmeq28w1FQqCZswHZ3xR+SuxYftzHq6WU6KJHvqxKzTxxQ==}
     dependencies:
-      '@types/d3-selection': 3.0.11
+      '@types/d3-selection': 3.0.10
     dev: false
 
   /@types/d3-interpolate@3.0.4:
@@ -3332,21 +3332,21 @@ packages:
       '@types/d3-color': 3.1.3
     dev: false
 
-  /@types/d3-selection@3.0.11:
-    resolution: {integrity: sha512-bhAXu23DJWsrI45xafYpkQ4NtcKMwWnAC/vKrd2l+nxMFuvOT3XMYTIj2opv8vq8AO5Yh7Qac/nSeP/3zjTK0w==}
+  /@types/d3-selection@3.0.10:
+    resolution: {integrity: sha512-cuHoUgS/V3hLdjJOLTT691+G2QoqAjCVLmr4kJXR4ha56w1Zdu8UUQ5TxLRqudgNjwXeQxKMq4j+lyf9sWuslg==}
     dev: false
 
-  /@types/d3-transition@3.0.9:
-    resolution: {integrity: sha512-uZS5shfxzO3rGlu0cC3bjmMFKsXv+SmZZcgp0KD22ts4uGXp5EVYGzu/0YdwZeKmddhcAccYtREJKkPfXkZuCg==}
+  /@types/d3-transition@3.0.8:
+    resolution: {integrity: sha512-ew63aJfQ/ms7QQ4X7pk5NxQ9fZH/z+i24ZfJ6tJSfqxJMrYLiK01EAs2/Rtw/JreGUsS3pLPNV644qXFGnoZNQ==}
     dependencies:
-      '@types/d3-selection': 3.0.11
+      '@types/d3-selection': 3.0.10
     dev: false
 
   /@types/d3-zoom@3.0.8:
     resolution: {integrity: sha512-iqMC4/YlFCSlO8+2Ii1GGGliCAY4XdeG748w5vQUbevlbDu0zSjH/+jojorQVBK/se0j6DUFNPBGSqD3YWYnDw==}
     dependencies:
       '@types/d3-interpolate': 3.0.4
-      '@types/d3-selection': 3.0.11
+      '@types/d3-selection': 3.0.10
     dev: false
 
   /@types/diff-match-patch@1.0.36:
@@ -3951,28 +3951,28 @@ packages:
     resolution: {integrity: sha512-N8tkAACJx2ww8vFMneJmaAgmjAG1tnVBZJRLRcx061tmsLRZHSEZSLuGWnwPtunsSLvSqXQ2wfp7Mgqg1I+2dQ==}
     dev: false
 
-  /@xyflow/react@12.5.3(@types/react@18.3.11)(react-dom@18.3.1)(react@18.3.1):
-    resolution: {integrity: sha512-saovy/aQRoW8qQoIqMFUtmC3F6oEV7n6+J1pVbhSG45NI/hOFvK0qozsIPKqX5Va6lGQnkl/o53NHLja3NiweQ==}
+  /@xyflow/react@12.4.2(@types/react@18.3.11)(react-dom@18.3.1)(react@18.3.1):
+    resolution: {integrity: sha512-AFJKVc/fCPtgSOnRst3xdYJwiEcUN9lDY7EO/YiRvFHYCJGgfzg+jpvZjkTOnBLGyrMJre9378pRxAc3fsR06A==}
     peerDependencies:
       react: '>=17'
       react-dom: '>=17'
     dependencies:
-      '@xyflow/system': 0.0.53
+      '@xyflow/system': 0.0.50
       classcat: 5.0.5
       react: 18.3.1
       react-dom: 18.3.1(react@18.3.1)
-      zustand: 4.5.6(@types/react@18.3.11)(react@18.3.1)
+      zustand: 4.5.5(@types/react@18.3.11)(react@18.3.1)
     transitivePeerDependencies:
       - '@types/react'
       - immer
     dev: false
 
-  /@xyflow/system@0.0.53:
-    resolution: {integrity: sha512-QTWieiTtvNYyQAz1fxpzgtUGXNpnhfh6vvZa7dFWpWS2KOz6bEHODo/DTK3s07lDu0Bq0Db5lx/5M5mNjb9VDQ==}
+  /@xyflow/system@0.0.50:
+    resolution: {integrity: sha512-HVUZd4LlY88XAaldFh2nwVxDOcdIBxGpQ5txzwfJPf+CAjj2BfYug1fHs2p4yS7YO8H6A3EFJQovBE8YuHkAdg==}
     dependencies:
       '@types/d3-drag': 3.0.7
-      '@types/d3-selection': 3.0.11
-      '@types/d3-transition': 3.0.9
+      '@types/d3-selection': 3.0.10
+      '@types/d3-transition': 3.0.8
       '@types/d3-zoom': 3.0.8
       d3-drag: 3.0.0
       d3-selection: 3.0.0
@@ -9123,14 +9123,6 @@ packages:
       react: 18.3.1
     dev: false
 
-  /use-sync-external-store@1.5.0(react@18.3.1):
-    resolution: {integrity: sha512-Rb46I4cGGVBmjamjphe8L/UnvJD+uPPtTkNvX5mZgqdbavhI4EbgIWJiIHXJ8bc/i9EQGPRh4DwEURJ552Do0A==}
-    peerDependencies:
-      react: ^16.8.0 || ^17.0.0 || ^18.0.0 || ^19.0.0
-    dependencies:
-      react: 18.3.1
-    dev: false
-
   /util-deprecate@1.0.2:
     resolution: {integrity: sha512-EPD5q1uXyFxJpCrLnCc1nHnq3gOa6DZBocAIiI2TaSCA7VCJ1UJDMagCzIkXNsUYfD1daK//LTEQ8xiIbrHtcw==}
     dev: true
@@ -9575,8 +9567,8 @@ packages:
   /zod@3.23.8:
     resolution: {integrity: sha512-XBx9AXhXktjUqnepgTiE5flcKIYWi/rme0Eaj+5Y0lftuGBq+jyRu/md4WnuxqgP1ubdpNCsYEYPxrzVHD8d6g==}
 
-  /zustand@4.5.6(@types/react@18.3.11)(react@18.3.1):
-    resolution: {integrity: sha512-ibr/n1hBzLLj5Y+yUcU7dYw8p6WnIVzdJbnX+1YpaScvZVF2ziugqHs+LAmHw4lWO9c/zRj+K1ncgWDQuthEdQ==}
+  /zustand@4.5.5(@types/react@18.3.11)(react@18.3.1):
+    resolution: {integrity: sha512-+0PALYNJNgK6hldkgDq2vLrw5f6g/jCInz52n9RTpropGgeAf/ioFUCdtsjCqu4gNhW9D01rUQBROoRjdzyn2Q==}
     engines: {node: '>=12.7.0'}
     peerDependencies:
       '@types/react': '>=16.8'
@@ -9592,5 +9584,5 @@ packages:
     dependencies:
       '@types/react': 18.3.11
       react: 18.3.1
-      use-sync-external-store: 1.5.0(react@18.3.1)
+      use-sync-external-store: 1.2.2(react@18.3.1)
     dev: false

invokeai/frontend/web/public/locales/de.json

@@ -116,10 +116,7 @@
         "combinatorial": "Kombinatorisch",
         "saveChanges": "Änderungen speichern",
         "error_withCount_one": "{{count}} Fehler",
-        "error_withCount_other": "{{count}} Fehler",
-        "value": "Wert",
-        "label": "Label",
-        "systemInformation": "Systeminformationen"
+        "error_withCount_other": "{{count}} Fehler"
     },
     "gallery": {
         "galleryImageSize": "Bildgröße",
@@ -698,10 +695,7 @@
         "guidance": "Führung",
         "coherenceMode": "Modus",
         "recallMetadata": "Metadaten abrufen",
-        "gaussianBlur": "Gaußsche Unschärfe",
-        "sendToUpscale": "An Hochskalieren senden",
-        "useCpuNoise": "CPU-Rauschen verwenden",
-        "sendToCanvas": "An Leinwand senden"
+        "gaussianBlur": "Gaußsche Unschärfe"
     },
     "settings": {
         "displayInProgress": "Zwischenbilder anzeigen",
@@ -1334,8 +1328,7 @@
         "loadWorkflowDesc2": "Ihr aktueller Arbeitsablauf enthält nicht gespeicherte Änderungen.",
         "loadingTemplates": "Lade {{name}}",
         "missingSourceOrTargetHandle": "Fehlender Quell- oder Zielgriff",
-        "missingSourceOrTargetNode": "Fehlender Quell- oder Zielknoten",
-        "showEdgeLabelsHelp": "Beschriftungen an Kanten anzeigen, um die verknüpften Knoten zu kennzeichnen"
+        "missingSourceOrTargetNode": "Fehlender Quell- oder Zielknoten"
     },
     "hrf": {
         "enableHrf": "Korrektur für hohe Auflösungen",

invokeai/frontend/web/public/locales/it.json

@@ -115,8 +115,7 @@
         "error_withCount_many": "{{count}} errori",
         "error_withCount_other": "{{count}} errori",
         "value": "Valore",
-        "label": "Etichetta",
-        "systemInformation": "Informazioni di sistema"
+        "label": "Etichetta"
     },
     "gallery": {
         "galleryImageSize": "Dimensione dell'immagine",
@@ -716,8 +715,7 @@
             "collectionNumberLTMin": "{{value}} < {{minimum}} (incr min)",
             "collectionNumberGTExclusiveMax": "{{value}} >= {{exclusiveMaximum}} (excl max)",
             "collectionNumberLTExclusiveMin": "{{value}} <= {{exclusiveMinimum}} (excl min)",
-            "collectionEmpty": "raccolta vuota",
-            "batchNodeCollectionSizeMismatchNoGroupId": "Dimensione della raccolta di gruppo nel Lotto non corrisponde"
+            "collectionEmpty": "raccolta vuota"
         },
         "useCpuNoise": "Usa la CPU per generare rumore",
         "iterations": "Iterazioni",
@@ -2367,9 +2365,8 @@
         "watchRecentReleaseVideos": "Guarda i video su questa versione",
         "watchUiUpdatesOverview": "Guarda le novità dell'interfaccia",
         "items": [
-            "Flussi di lavoro: supporto per menu a discesa di stringhe personalizzate nel Generatore di Flussi di lavoro.",
-            "FLUX: supporto per FLUX Fill in Flussi di lavoro e Tela.",
-            "LLaVA OneVision VLLM: supporto beta nei flussi di lavoro."
+            "Flussi di lavoro: nuova e migliorata libreria dei flussi di lavoro.",
+            "FLUX: supporto per FLUX Redux e FLUX Fill in Flussi di lavoro e Tela."
         ]
     },
     "system": {

invokeai/frontend/web/public/locales/vi.json

@@ -237,10 +237,7 @@
         "row": "Hàng",
         "board": "Bảng",
         "saveChanges": "Lưu Thay Đổi",
-        "error_withCount_other": "{{count}} lỗi",
-        "value": "Giá Trị",
-        "label": "Nhãn Tên",
-        "systemInformation": "Thông Tin Hệ Thống"
+        "error_withCount_other": "{{count}} lỗi"
     },
     "prompt": {
         "addPromptTrigger": "Thêm Prompt Trigger",
@@ -2303,10 +2300,7 @@
             "minimum": "Tối Thiểu",
             "maximum": "Tối Đa",
             "containerRowLayout": "Hộp Chứa (bố cục hàng)",
-            "containerColumnLayout": "Hộp Chứa (bố cục cột)",
-            "resetOptions": "Tải Lại Lựa Chọn",
-            "addOption": "Thêm Lựa Chọn",
-            "dropdown": "Danh Sách Thả Xuống"
+            "containerColumnLayout": "Hộp Chứa (bố cục cột)"
         },
         "yourWorkflows": "Workflow Của Bạn",
         "browseWorkflows": "Khám Phá Workflow",
@@ -2322,8 +2316,7 @@
         "view": "Xem",
         "deselectAll": "Huỷ Chọn Tất Cả",
         "noRecentWorkflows": "Không Có Workflows Gần Đây",
-        "recommended": "Có Thể Bạn Sẽ Cần",
-        "emptyStringPlaceholder": "<xâu ký tự trống>"
+        "recommended": "Có Thể Bạn Sẽ Cần"
     },
     "upscaling": {
         "missingUpscaleInitialImage": "Thiếu ảnh dùng để upscale",
@@ -2359,9 +2352,8 @@
         "watchRecentReleaseVideos": "Xem Video Phát Hành Mới Nhất",
         "watchUiUpdatesOverview": "Xem Tổng Quan Về Những Cập Nhật Cho Giao Diện Người Dùng",
         "items": [
-            "Workflow: Hỗ trợ xâu ký tự thả xuống tùy chỉnh trong Trình Tạo Vùng Nhập.",
-            "FLUX: Hỗ trợ FLUX Fill trong Workflow và Canvas.",
-            "LLaVA OneVision VLLM: Hỗ trợ phiên bản Beta trong Workflow."
+            "Workflow: Thư Viện Workflow mới và đã được cải tiến.",
+            "FLUX: Hỗ trợ FLUX Redux & FLUX Fill trong Workflow và Canvas."
         ]
     },
     "upsell": {

invokeai/frontend/web/src/common/hooks/useClientSideUpload.ts

@@ -6,16 +6,6 @@ import { selectAutoAddBoardId } from 'features/gallery/store/gallerySelectors';
 import { useCallback } from 'react';
 import { useCreateImageUploadEntryMutation } from 'services/api/endpoints/images';
 import type { ImageDTO } from 'services/api/types';
-
-type PresignedUrlResponse = {
-  fullUrl: string;
-  thumbnailUrl: string;
-};
-
-const isPresignedUrlResponse = (response: unknown): response is PresignedUrlResponse => {
-  return typeof response === 'object' && response !== null && 'fullUrl' in response && 'thumbnailUrl' in response;
-};
-
 export const useClientSideUpload = () => {
   const dispatch = useAppDispatch();
   const autoAddBoardId = useAppSelector(selectAutoAddBoardId);
@@ -84,30 +74,24 @@ export const useClientSideUpload = () => {
         board_id: autoAddBoardId === 'none' ? undefined : autoAddBoardId,
       }).unwrap();
 
-      const response = await fetch(presigned_url, {
-        method: 'GET',
+      await fetch(`${presigned_url}/?type=full`, {
+        method: 'PUT',
+        body: file,
         ...(authToken && {
           headers: {
             Authorization: `Bearer ${authToken}`,
           },
         }),
-      }).then((res) => res.json());
-
-      if (!isPresignedUrlResponse(response)) {
-        throw new Error('Invalid response');
-      }
-
-      const fullUrl = response.fullUrl;
-      const thumbnailUrl = response.thumbnailUrl;
-
-      await fetch(fullUrl, {
-        method: 'PUT',
-        body: file,
       });
 
-      await fetch(thumbnailUrl, {
+      await fetch(`${presigned_url}/?type=thumbnail`, {
         method: 'PUT',
         body: thumbnail,
+        ...(authToken && {
+          headers: {
+            Authorization: `Bearer ${authToken}`,
+          },
+        }),
       });
 
       dispatch(imageUploadedClientSide({ imageDTO: image_dto, silent: false, isFirstUploadOfBatch: i === 0 }));

invokeai/frontend/web/src/common/hooks/useImageUploadButton.tsx

@@ -58,58 +58,50 @@ export const useImageUploadButton = ({ onUpload, isDisabled, allowMultiple }: Us
 
   const onDropAccepted = useCallback(
     async (files: File[]) => {
-      try {
-        if (!allowMultiple) {
-          if (files.length > 1) {
-            log.warn('Multiple files dropped but only one allowed');
-            return;
-          }
-          if (files.length === 0) {
-            // Should never happen
-            log.warn('No files dropped');
-            return;
-          }
-          const file = files[0];
-          assert(file !== undefined); // should never happen
-          const imageDTO = await uploadImage({
-            file,
-            image_category: 'user',
-            is_intermediate: false,
-            board_id: autoAddBoardId === 'none' ? undefined : autoAddBoardId,
-            silent: true,
-          }).unwrap();
-          if (onUpload) {
-            onUpload(imageDTO);
-          }
-        } else {
-          let imageDTOs: ImageDTO[] = [];
-          if (isClientSideUploadEnabled) {
-            imageDTOs = await Promise.all(files.map((file, i) => clientSideUpload(file, i)));
-          } else {
-            imageDTOs = await uploadImages(
-              files.map((file, i) => ({
-                file,
-                image_category: 'user',
-                is_intermediate: false,
-                board_id: autoAddBoardId === 'none' ? undefined : autoAddBoardId,
-                silent: false,
-                isFirstUploadOfBatch: i === 0,
-              }))
-            );
-          }
-          if (onUpload) {
-            onUpload(imageDTOs);
-          }
+      if (!allowMultiple) {
+        if (files.length > 1) {
+          log.warn('Multiple files dropped but only one allowed');
+          return;
+        }
+        if (files.length === 0) {
+          // Should never happen
+          log.warn('No files dropped');
+          return;
+        }
+        const file = files[0];
+        assert(file !== undefined); // should never happen
+        const imageDTO = await uploadImage({
+          file,
+          image_category: 'user',
+          is_intermediate: false,
+          board_id: autoAddBoardId === 'none' ? undefined : autoAddBoardId,
+          silent: true,
+        }).unwrap();
+        if (onUpload) {
+          onUpload(imageDTO);
+        }
+      } else {
+        let imageDTOs: ImageDTO[] = [];
+        if (isClientSideUploadEnabled) {
+          imageDTOs = await Promise.all(files.map((file, i) => clientSideUpload(file, i)));
+        } else {
+          imageDTOs = await uploadImages(
+            files.map((file, i) => ({
+              file,
+              image_category: 'user',
+              is_intermediate: false,
+              board_id: autoAddBoardId === 'none' ? undefined : autoAddBoardId,
+              silent: false,
+              isFirstUploadOfBatch: i === 0,
+            }))
+          );
+        }
+        if (onUpload) {
+          onUpload(imageDTOs);
         }
-      } catch (error) {
-        toast({
-          id: 'UPLOAD_FAILED',
-          title: t('toast.imageUploadFailed'),
-          status: 'error',
-        });
       }
     },
-    [allowMultiple, autoAddBoardId, onUpload, uploadImage, isClientSideUploadEnabled, clientSideUpload, t]
+    [allowMultiple, autoAddBoardId, onUpload, uploadImage, isClientSideUploadEnabled, clientSideUpload]
   );
 
   const onDropRejected = useCallback(

invokeai/frontend/web/src/features/controlLayers/components/ParamDenoisingStrength.tsx

@@ -14,9 +14,8 @@ import WavyLine from 'common/components/WavyLine';
 import { selectImg2imgStrength, setImg2imgStrength } from 'features/controlLayers/store/paramsSlice';
 import { selectActiveRasterLayerEntities } from 'features/controlLayers/store/selectors';
 import { selectImg2imgStrengthConfig } from 'features/system/store/configSlice';
-import { memo, useCallback, useMemo } from 'react';
+import { memo, useCallback } from 'react';
 import { useTranslation } from 'react-i18next';
-import { useSelectedModelConfig } from 'services/api/hooks/useSelectedModelConfig';
 
 const selectHasRasterLayersWithContent = createSelector(
   selectActiveRasterLayerEntities,
@@ -27,7 +26,6 @@ export const ParamDenoisingStrength = memo(() => {
   const img2imgStrength = useAppSelector(selectImg2imgStrength);
   const dispatch = useAppDispatch();
   const hasRasterLayersWithContent = useAppSelector(selectHasRasterLayersWithContent);
-  const selectedModelConfig = useSelectedModelConfig();
 
   const onChange = useCallback(
     (v: number) => {
@@ -41,24 +39,8 @@ export const ParamDenoisingStrength = memo(() => {
 
   const [invokeBlue300] = useToken('colors', ['invokeBlue.300']);
 
-  const isDisabled = useMemo(() => {
-    if (!hasRasterLayersWithContent) {
-      // Denoising strength does nothing if there are no raster layers w/ content
-      return true;
-    }
-    if (
-      selectedModelConfig?.type === 'main' &&
-      selectedModelConfig?.base === 'flux' &&
-      selectedModelConfig.variant === 'inpaint'
-    ) {
-      // Denoising strength is ignored by FLUX Fill, which is indicated by the variant being 'inpaint'
-      return true;
-    }
-    return false;
-  }, [hasRasterLayersWithContent, selectedModelConfig]);
-
   return (
-    <FormControl isDisabled={isDisabled} p={1} justifyContent="space-between" h={8}>
+    <FormControl isDisabled={!hasRasterLayersWithContent} p={1} justifyContent="space-between" h={8}>
       <Flex gap={3} alignItems="center">
         <InformationalPopover feature="paramDenoisingStrength">
           <FormLabel mr={0}>{`${t('parameters.denoisingStrength')}`}</FormLabel>
@@ -67,7 +49,7 @@ export const ParamDenoisingStrength = memo(() => {
           <WavyLine amplitude={img2imgStrength * 10} stroke={invokeBlue300} strokeWidth={1} width={40} height={14} />
         )}
       </Flex>
-      {!isDisabled ? (
+      {hasRasterLayersWithContent ? (
         <>
           <CompositeSlider
             step={config.coarseStep}

invokeai/frontend/web/src/features/gallery/hooks/useGalleryHotkeys.ts

@@ -49,11 +49,7 @@ export const useGalleryHotkeys = () => {
   useRegisteredHotkeys({
     id: 'galleryNavLeft',
     category: 'gallery',
-    callback: (e) => {
-      // Skip the hotkey if the user is focused on a tab element - the arrow keys are used to navigate between tabs.
-      if (e.target instanceof HTMLElement && e.target.getAttribute('role') === 'tab') {
-        return;
-      }
+    callback: () => {
       if (isOnFirstImageOfView && isPrevEnabled && !queryResult.isFetching) {
         goPrev('arrow');
         return;
@@ -75,11 +71,7 @@ export const useGalleryHotkeys = () => {
   useRegisteredHotkeys({
     id: 'galleryNavRight',
     category: 'gallery',
-    callback: (e) => {
-      // Skip the hotkey if the user is focused on a tab element - the arrow keys are used to navigate between tabs.
-      if (e.target instanceof HTMLElement && e.target.getAttribute('role') === 'tab') {
-        return;
-      }
+    callback: () => {
       if (isOnLastImageOfView && isNextEnabled && !queryResult.isFetching) {
         goNext('arrow');
         return;

invokeai/frontend/web/src/features/nodes/store/nodesSlice.ts

@@ -470,8 +470,31 @@ export const nodesSlice = createSlice({
     builder.addCase(workflowLoaded, (state, action) => {
       const { nodes, edges } = action.payload;
 
-      state.nodes = nodes.map((node) => ({ ...SHARED_NODE_PROPERTIES, ...node }));
-      state.edges = edges;
+      const changes: NodeChange<AnyNode>[] = [];
+      for (const node of nodes) {
+        if (node.type === 'notes') {
+          changes.push({
+            type: 'add',
+            item: {
+              ...SHARED_NODE_PROPERTIES,
+              ...node,
+            },
+          });
+        } else if (node.type === 'invocation') {
+          changes.push({
+            type: 'add',
+            item: {
+              ...SHARED_NODE_PROPERTIES,
+              ...node,
+            },
+          });
+        }
+      }
+      state.nodes = applyNodeChanges<AnyNode>(changes, []);
+      state.edges = applyEdgeChanges(
+        edges.map((edge) => ({ type: 'add', item: edge })),
+        []
+      );
     });
   },
 });

invokeai/frontend/web/src/features/nodes/types/openapi.ts

@@ -79,4 +79,4 @@ export const isInvocationOutputSchemaObject = (
 
 export const isInvocationFieldSchema = (
   obj: OpenAPIV3_1.ReferenceObject | OpenAPIV3_1.SchemaObject
-): obj is InvocationFieldSchema => 'field_kind' in obj;
+): obj is InvocationFieldSchema => !('$ref' in obj);

invokeai/frontend/web/src/features/nodes/util/workflow/validateWorkflow.ts

@@ -148,11 +148,7 @@ export const validateWorkflow = async (args: ValidateWorkflowArgs): Promise<Vali
       }
     }
   }
-
-  // Stash invalid edges here to be deleted later
-  const edgesToDelete = new Set<string>();
-
-  for (const edge of edges) {
+  edges.forEach((edge, i) => {
     // Validate each edge. If the edge is invalid, we must remove it to prevent runtime errors with reactflow.
     const sourceNode = nodes.find(({ id }) => id === edge.source);
     const targetNode = nodes.find(({ id }) => id === edge.target);
@@ -219,7 +215,8 @@ export const validateWorkflow = async (args: ValidateWorkflowArgs): Promise<Vali
     }
 
     if (issues.length) {
-      edgesToDelete.add(edge.id);
+      // This edge has some issues. Remove it.
+      delete edges[i];
       const source = edge.type === 'default' ? `${edge.source}.${edge.sourceHandle}` : edge.source;
       const target = edge.type === 'default' ? `${edge.source}.${edge.targetHandle}` : edge.target;
       warnings.push({
@@ -228,10 +225,7 @@ export const validateWorkflow = async (args: ValidateWorkflowArgs): Promise<Vali
         data: edge,
       });
     }
-  }
-
-  // Remove invalid edges
-  _workflow.edges = edges.filter(({ id }) => !edgesToDelete.has(id));
+  });
 
   // Migrated exposed fields to form elements if they exist and the form does not
   // Note: If the form is invalid per its zod schema, it will be reset to a default, empty form!

invokeai/version/invokeai_version.py

@@ -1 +1 @@
-__version__ = "5.10.0dev1"
+__version__ = "5.9.0"

pins.json

@@ -1,14 +0,0 @@
-{
-  "python": "3.12",
-  "torchIndexUrl": {
-    "win32": {
-      "cuda": "https://download.pytorch.org/whl/cu126"
-    },
-    "linux": {
-      "cpu": "https://download.pytorch.org/whl/cpu",
-      "rocm": "https://download.pytorch.org/whl/rocm6.2.4",
-      "cuda": "https://download.pytorch.org/whl/cu126"
-    },
-    "darwin": {}
-  }
-}

pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "setuptools.build_meta"
 [project]
 name = "InvokeAI"
 description = "An implementation of Stable Diffusion which provides various new features and options to aid the image generation process"
-requires-python = ">=3.10, <3.13"
+requires-python = ">=3.10, <3.12"
 readme = { content-type = "text/markdown", file = "README.md" }
 keywords = ["stable-diffusion", "AI"]
 dynamic = ["version"]
@@ -33,12 +33,12 @@ classifiers = [
 ]
 dependencies = [
   # Core generation dependencies, pinned for reproducible builds.
-  "accelerate",
-  "bitsandbytes; sys_platform!='darwin'",
+  "accelerate==1.0.1",
+  "bitsandbytes==0.45.0; sys_platform!='darwin'",
   "clip_anytorch==2.6.0",       # replacing "clip @ https://github.com/openai/CLIP/archive/eaa22acb90a5876642d0507623e859909230a52d.zip",
   "compel==2.0.2",
   "controlnet-aux==0.0.7",
-  "diffusers[torch]",
+  "diffusers[torch]==0.31.0",
   "gguf==0.10.0",
   "invisible-watermark==0.2.0", # needed to install SDXL base and refiner using their repo_ids
   "mediapipe==0.10.14",         # needed for "mediapipeface" controlnet model
@@ -46,27 +46,26 @@ dependencies = [
   "onnx==1.16.1",
   "onnxruntime==1.19.2",
   "opencv-python==4.9.0.80",
-  "pytorch-lightning",
-  "safetensors",
+  "pytorch-lightning==2.1.3",
+  "safetensors==0.4.3",
   # sentencepiece is required to load T5TokenizerFast (used by FLUX).
   "sentencepiece==0.2.0",
   "spandrel==0.3.4",
-  "timm~=1.0.0",
-  "torch~=2.6.0",                # torch and related dependencies are loosely pinned, will respect requirement of `diffusers[torch]`
+  "timm==0.6.13",               # needed to override timm latest in controlnet_aux, see  https://github.com/isl-org/ZoeDepth/issues/26
+  "torch<2.5.0",                # torch and related dependencies are loosely pinned, will respect requirement of `diffusers[torch]`
   "torchmetrics",
   "torchsde",
   "torchvision",
-  "timm~=1.0.0",                 # controlnet-aux depends on a version of timm that breaks LLaVA. explicitly pinning `timm` here, which results in a downgrade of `controlnet-aux`. https://github.com/huggingface/controlnet_aux/pull/101. If this poses a problem, we need to decide between newer `controlnet_aux` and LLaVA, OR we can fork `controlnet_aux` and update the pin.
-  "transformers",
+  "transformers==4.46.3",
 
   # Core application dependencies, pinned for reproducible builds.
-  "fastapi-events",
-  "fastapi",
-  "huggingface-hub",
-  "pydantic-settings",
-  "pydantic",
-  "python-socketio",
-  "uvicorn[standard]",
+  "fastapi-events==0.11.1",
+  "fastapi==0.111.0",
+  "huggingface-hub==0.26.1",
+  "pydantic-settings==2.2.1",
+  "pydantic==2.7.2",
+  "python-socketio==5.11.1",
+  "uvicorn[standard]==0.28.0",
 
   # Auxiliary dependencies, pinned only if necessary.
   "albumentations",
@@ -91,8 +90,11 @@ dependencies = [
   "pyreadline3",
   "python-multipart",
   "requests",
+  "rich~=13.3",
   "scikit-image",
   "semver~=3.0.1",
+  "test-tube",
+  "windows-curses; sys_platform=='win32'",
   "humanize==4.12.1",
 ]
 
@@ -115,7 +117,7 @@ dependencies = [
 ]
 "dev" = ["jurigged", "pudb", "snakeviz", "gprof2dot"]
 "test" = [
-  "ruff~=0.11.2",
+  "ruff~=0.9.9",
   "ruff-lsp~=0.0.62",
   "mypy",
   "pre-commit",
@@ -142,7 +144,19 @@ dependencies = [
 version = { attr = "invokeai.version.__version__" }
 
 [tool.setuptools.packages.find]
-"include" = ["invokeai"]
+"where" = ["."]
+"include" = [
+  "invokeai.assets.fonts*",
+  "invokeai.version*",
+  "invokeai.generator*",
+  "invokeai.backend*",
+  "invokeai.frontend*",
+  "invokeai.frontend.web.dist*",
+  "invokeai.frontend.web.static*",
+  "invokeai.configs*",
+  "invokeai.app*",
+  "invokeai.invocation_api*",
+]
 
 [tool.setuptools.package-data]
 "invokeai.app.assets" = ["**/*.png"]

scripts/strip_models.py

@@ -22,8 +22,9 @@ from pathlib import Path
 import humanize
 import torch
 
-from invokeai.backend.model_manager.model_on_disk import ModelOnDisk
+from invokeai.backend.model_manager.config import ModelOnDisk
 from invokeai.backend.model_manager.search import ModelSearch
+from invokeai.backend.model_manager.taxonomy import ModelFormat
 
 
 def strip(v):
@@ -62,7 +63,7 @@ def load_stripped_model(path: Path, *args, **kwargs):
 
 def create_stripped_model(original_model_path: Path, stripped_model_path: Path) -> ModelOnDisk:
     original = ModelOnDisk(original_model_path)
-    if original.path.is_file():
+    if original.format_type == ModelFormat.Checkpoint:
         shutil.copy2(original.path, stripped_model_path)
     else:
         shutil.copytree(original.path, stripped_model_path, dirs_exist_ok=True)

tests/backend/flux/controlnet/test_state_dict_utils.py

@@ -24,7 +24,7 @@ from tests.backend.flux.controlnet.xlabs_flux_controlnet_state_dict import xlabs
     ],
 )
 def test_is_state_dict_xlabs_controlnet(sd_shapes: dict[str, list[int]], expected: bool):
-    sd = dict.fromkeys(sd_shapes)
+    sd = {k: None for k in sd_shapes}
     assert is_state_dict_xlabs_controlnet(sd) == expected
 
 
@@ -37,7 +37,7 @@ def test_is_state_dict_xlabs_controlnet(sd_shapes: dict[str, list[int]], expecte
     ],
 )
 def test_is_state_dict_instantx_controlnet(sd_keys: list[str], expected: bool):
-    sd = dict.fromkeys(sd_keys)
+    sd = {k: None for k in sd_keys}
     assert is_state_dict_instantx_controlnet(sd) == expected
 
 

tests/backend/flux/ip_adapter/test_xlabs_ip_adapter_flux.py

@@ -19,7 +19,7 @@ from tests.backend.flux.ip_adapter.xlabs_flux_ip_adapter_v2_state_dict import xl
 @pytest.mark.parametrize("sd_shapes", [xlabs_flux_ip_adapter_sd_shapes, xlabs_flux_ip_adapter_v2_sd_shapes])
 def test_is_state_dict_xlabs_ip_adapter(sd_shapes: dict[str, list[int]]):
     # Construct a dummy state_dict.
-    sd = dict.fromkeys(sd_shapes)
+    sd = {k: None for k in sd_shapes}
 
     assert is_state_dict_xlabs_ip_adapter(sd)
 

tests/test_config.py

@@ -5,7 +5,7 @@ from typing import Any
 import pytest
 from pydantic import ValidationError
 
-from invokeai.app.invocations.baseinvocation import InvocationRegistry
+from invokeai.app.invocations.baseinvocation import BaseInvocation
 from invokeai.app.services.config.config_default import (
     DefaultInvokeAIAppConfig,
     InvokeAIAppConfig,
@@ -274,7 +274,7 @@ def test_deny_nodes(patch_rootdir):
 
     # We've changed the config, we need to invalidate the typeadapter cache so that the new config is used for
     # subsequent graph validations
-    InvocationRegistry.invalidate_invocation_typeadapter()
+    BaseInvocation.invalidate_typeadapter()
 
     # confirm graph validation fails when using denied node
     Graph.model_validate({"nodes": {"1": {"id": "1", "type": "integer"}}})
@@ -284,7 +284,7 @@ def test_deny_nodes(patch_rootdir):
         Graph.model_validate({"nodes": {"1": {"id": "1", "type": "float"}}})
 
     # confirm invocations union will not have denied nodes
-    all_invocations = InvocationRegistry.get_invocation_classes()
+    all_invocations = BaseInvocation.get_invocations()
 
     has_integer = len([i for i in all_invocations if i.get_type() == "integer"]) == 1
     has_string = len([i for i in all_invocations if i.get_type() == "string"]) == 1
@@ -296,4 +296,4 @@ def test_deny_nodes(patch_rootdir):
 
     # Reset the config so that it doesn't affect other tests
     get_config.cache_clear()
-    InvocationRegistry.invalidate_invocation_typeadapter()
+    BaseInvocation.invalidate_typeadapter()

tests/test_model_probe.py

@@ -17,6 +17,7 @@ from invokeai.backend.model_manager.config import (
     MainDiffusersConfig,
     ModelConfigBase,
     ModelConfigFactory,
+    ModelOnDisk,
     get_model_discriminator_value,
 )
 from invokeai.backend.model_manager.legacy_probe import (
@@ -26,7 +27,6 @@ from invokeai.backend.model_manager.legacy_probe import (
     get_default_settings_control_adapters,
     get_default_settings_main,
 )
-from invokeai.backend.model_manager.model_on_disk import ModelOnDisk
 from invokeai.backend.model_manager.search import ModelSearch
 from invokeai.backend.util.logging import InvokeAILogger
 

tests/test_model_probe/stripped_models/Hard Edge Detection (Canny).safetensors

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5acefb3658338a4126736e2da02cfef5a9ce6e2469564a6c7994ae34e8ef2e8a
-size 192447

tests/test_model_probe/stripped_models/flat_colour_anime_style_schnell_v3.4.safetensors

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:55aafd0f9b4ac2863361573b070320e13b800b2359a81a73878008bdffc3edfa
-size 201040

tests/test_object_serializer_disk.py

@@ -21,18 +21,16 @@ def count_files(path: Path):
 
 @pytest.fixture
 def obj_serializer(tmp_path: Path):
-    return ObjectSerializerDisk[MockDataclass](tmp_path, safe_globals=[MockDataclass])
+    return ObjectSerializerDisk[MockDataclass](tmp_path)
 
 
 @pytest.fixture
 def fwd_cache(tmp_path: Path):
-    return ObjectSerializerForwardCache(
-        ObjectSerializerDisk[MockDataclass](tmp_path, safe_globals=[MockDataclass]), max_cache_size=2
-    )
+    return ObjectSerializerForwardCache(ObjectSerializerDisk[MockDataclass](tmp_path), max_cache_size=2)
 
 
 def test_obj_serializer_disk_initializes(tmp_path: Path):
-    obj_serializer = ObjectSerializerDisk[MockDataclass](tmp_path, safe_globals=[MockDataclass])
+    obj_serializer = ObjectSerializerDisk[MockDataclass](tmp_path)
     assert obj_serializer._output_dir == tmp_path
 
 
@@ -72,7 +70,7 @@ def test_obj_serializer_disk_deletes(obj_serializer: ObjectSerializerDisk[MockDa
 
 
 def test_obj_serializer_ephemeral_creates_tempdir(tmp_path: Path):
-    obj_serializer = ObjectSerializerDisk[MockDataclass](tmp_path, safe_globals=[MockDataclass], ephemeral=True)
+    obj_serializer = ObjectSerializerDisk[MockDataclass](tmp_path, ephemeral=True)
     assert isinstance(obj_serializer._tempdir, tempfile.TemporaryDirectory)
     assert obj_serializer._base_output_dir == tmp_path
     assert obj_serializer._output_dir != tmp_path
@@ -80,21 +78,21 @@ def test_obj_serializer_ephemeral_creates_tempdir(tmp_path: Path):
 
 
 def test_obj_serializer_ephemeral_deletes_tempdir(tmp_path: Path):
-    obj_serializer = ObjectSerializerDisk[MockDataclass](tmp_path, safe_globals=[MockDataclass], ephemeral=True)
+    obj_serializer = ObjectSerializerDisk[MockDataclass](tmp_path, ephemeral=True)
     tempdir_path = obj_serializer._output_dir
     del obj_serializer
     assert not tempdir_path.exists()
 
 
 def test_obj_serializer_ephemeral_deletes_tempdir_on_stop(tmp_path: Path):
-    obj_serializer = ObjectSerializerDisk[MockDataclass](tmp_path, safe_globals=[MockDataclass], ephemeral=True)
+    obj_serializer = ObjectSerializerDisk[MockDataclass](tmp_path, ephemeral=True)
     tempdir_path = obj_serializer._output_dir
     obj_serializer.stop(None)  # pyright: ignore [reportArgumentType]
     assert not tempdir_path.exists()
 
 
 def test_obj_serializer_ephemeral_writes_to_tempdir(tmp_path: Path):
-    obj_serializer = ObjectSerializerDisk[MockDataclass](tmp_path, safe_globals=[MockDataclass], ephemeral=True)
+    obj_serializer = ObjectSerializerDisk[MockDataclass](tmp_path, ephemeral=True)
     obj_1 = MockDataclass(foo="bar")
     obj_1_name = obj_serializer.save(obj_1)
     assert Path(obj_serializer._output_dir, obj_1_name).exists()
@@ -104,19 +102,19 @@ def test_obj_serializer_ephemeral_writes_to_tempdir(tmp_path: Path):
 def test_obj_serializer_ephemeral_deletes_dangling_tempdirs_on_init(tmp_path: Path):
     tempdir = tmp_path / "tmpdir"
     tempdir.mkdir()
-    ObjectSerializerDisk[MockDataclass](tmp_path, safe_globals=[MockDataclass], ephemeral=True)
+    ObjectSerializerDisk[MockDataclass](tmp_path, ephemeral=True)
     assert not tempdir.exists()
 
 
 def test_obj_serializer_does_not_delete_tempdirs_on_init(tmp_path: Path):
     tempdir = tmp_path / "tmpdir"
     tempdir.mkdir()
-    ObjectSerializerDisk[MockDataclass](tmp_path, safe_globals=[MockDataclass], ephemeral=False)
+    ObjectSerializerDisk[MockDataclass](tmp_path, ephemeral=False)
     assert tempdir.exists()
 
 
 def test_obj_serializer_disk_different_types(tmp_path: Path):
-    obj_serializer_1 = ObjectSerializerDisk[MockDataclass](tmp_path, safe_globals=[MockDataclass])
+    obj_serializer_1 = ObjectSerializerDisk[MockDataclass](tmp_path)
     obj_1 = MockDataclass(foo="bar")
     obj_1_name = obj_serializer_1.save(obj_1)
     obj_1_loaded = obj_serializer_1.load(obj_1_name)
@@ -125,19 +123,19 @@ def test_obj_serializer_disk_different_types(tmp_path: Path):
     assert obj_1_loaded.foo == "bar"
     assert obj_1_name.startswith("MockDataclass_")
 
-    obj_serializer_2 = ObjectSerializerDisk[int](tmp_path, safe_globals=[int])
+    obj_serializer_2 = ObjectSerializerDisk[int](tmp_path)
     obj_2_name = obj_serializer_2.save(9001)
     assert obj_serializer_2._obj_class_name == "int"
     assert obj_serializer_2.load(obj_2_name) == 9001
     assert obj_2_name.startswith("int_")
 
-    obj_serializer_3 = ObjectSerializerDisk[str](tmp_path, safe_globals=[str])
+    obj_serializer_3 = ObjectSerializerDisk[str](tmp_path)
     obj_3_name = obj_serializer_3.save("foo")
     assert obj_serializer_3._obj_class_name == "str"
     assert obj_serializer_3.load(obj_3_name) == "foo"
     assert obj_3_name.startswith("str_")
 
-    obj_serializer_4 = ObjectSerializerDisk[torch.Tensor](tmp_path, safe_globals=[torch.Tensor])
+    obj_serializer_4 = ObjectSerializerDisk[torch.Tensor](tmp_path)
     obj_4_name = obj_serializer_4.save(torch.tensor([1, 2, 3]))
     obj_4_loaded = obj_serializer_4.load(obj_4_name)
     assert obj_serializer_4._obj_class_name == "Tensor"