github/InvokeAI
1
1
Fork 0
mirror of https://github.com/invoke-ai/InvokeAI.git synced 2026-02-04 22:15:08 -05:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'main' into refactor/rename-get-logger

Browse Source
This commit is contained in:
psychedelicious
2023-09-05 10:37:53 +10:00
committed by GitHub
parent 24132a7950 78cc5a7825
commit d8f7c19030
232 changed files with 8608 additions and 11111 deletions
Download Patch File Download Diff File Expand all files Collapse all files

216
invokeai/app/invocations/baseinvocation.py
View File

@@ -2,15 +2,18 @@
from __future__ import annotations
import json
from abc import ABC, abstractmethod
from enum import Enum
from inspect import signature
import re
from typing import (
TYPE_CHECKING,
AbstractSet,
Any,
Callable,
ClassVar,
Literal,
Mapping,
Optional,
Type,
@@ -20,14 +23,19 @@ from typing import (
get_type_hints,
)
from pydantic import BaseModel, Field
from pydantic.fields import Undefined
from pydantic import BaseModel, Field, validator
from pydantic.fields import Undefined, ModelField
from pydantic.typing import NoArgAnyCallable
import semver
if TYPE_CHECKING:
from ..services.invocation_services import InvocationServices
class InvalidVersionError(ValueError):
pass
class FieldDescriptions:
denoising_start = "When to start denoising, expressed a percentage of total steps"
denoising_end = "When to stop denoising, expressed a percentage of total steps"
@@ -102,24 +110,39 @@ class UIType(str, Enum):
"""
# region Primitives
Integer = "integer"
Float = "float"
Boolean = "boolean"
String = "string"
Array = "array"
Image = "ImageField"
Latents = "LatentsField"
Color = "ColorField"
Conditioning = "ConditioningField"
Control = "ControlField"
Color = "ColorField"
ImageCollection = "ImageCollection"
ConditioningCollection = "ConditioningCollection"
ColorCollection = "ColorCollection"
LatentsCollection = "LatentsCollection"
IntegerCollection = "IntegerCollection"
FloatCollection = "FloatCollection"
StringCollection = "StringCollection"
Float = "float"
Image = "ImageField"
Integer = "integer"
Latents = "LatentsField"
String = "string"
# endregion
# region Collection Primitives
BooleanCollection = "BooleanCollection"
ColorCollection = "ColorCollection"
ConditioningCollection = "ConditioningCollection"
ControlCollection = "ControlCollection"
FloatCollection = "FloatCollection"
ImageCollection = "ImageCollection"
IntegerCollection = "IntegerCollection"
LatentsCollection = "LatentsCollection"
StringCollection = "StringCollection"
# endregion
# region Polymorphic Primitives
BooleanPolymorphic = "BooleanPolymorphic"
ColorPolymorphic = "ColorPolymorphic"
ConditioningPolymorphic = "ConditioningPolymorphic"
ControlPolymorphic = "ControlPolymorphic"
FloatPolymorphic = "FloatPolymorphic"
ImagePolymorphic = "ImagePolymorphic"
IntegerPolymorphic = "IntegerPolymorphic"
LatentsPolymorphic = "LatentsPolymorphic"
StringPolymorphic = "StringPolymorphic"
# endregion
# region Models
@@ -141,9 +164,11 @@ class UIType(str, Enum):
# endregion
# region Misc
FilePath = "FilePath"
Enum = "enum"
Scheduler = "Scheduler"
WorkflowField = "WorkflowField"
IsIntermediate = "IsIntermediate"
MetadataField = "MetadataField"
# endregion
@@ -171,6 +196,7 @@ class _InputField(BaseModel):
ui_type: Optional[UIType]
ui_component: Optional[UIComponent]
ui_order: Optional[int]
item_default: Optional[Any]
class _OutputField(BaseModel):
@@ -218,6 +244,7 @@ def InputField(
ui_component: Optional[UIComponent] = None,
ui_hidden: bool = False,
ui_order: Optional[int] = None,
item_default: Optional[Any] = None,
**kwargs: Any,
) -> Any:
"""
@@ -244,6 +271,11 @@ def InputField(
For this case, you could provide `UIComponent.Textarea`.
: param bool ui_hidden: [False] Specifies whether or not this field should be hidden in the UI.
: param int ui_order: [None] Specifies the order in which this field should be rendered in the UI. \
: param bool item_default: [None] Specifies the default item value, if this is a collection input. \
Ignored for non-collection fields..
"""
return Field(
*args,
@@ -277,6 +309,7 @@ def InputField(
ui_component=ui_component,
ui_hidden=ui_hidden,
ui_order=ui_order,
item_default=item_default,
**kwargs,
)
@@ -327,6 +360,8 @@ def OutputField(
`UIType.SDXLMainModelField` to indicate that the field is an SDXL main model field.
: param bool ui_hidden: [False] Specifies whether or not this field should be hidden in the UI. \
: param int ui_order: [None] Specifies the order in which this field should be rendered in the UI. \
"""
return Field(
*args,
@@ -365,12 +400,15 @@ def OutputField(
class UIConfigBase(BaseModel):
"""
Provides additional node configuration to the UI.
This is used internally by the @tags and @title decorator logic. You probably want to use those
decorators, though you may add this class to a node definition to specify the title and tags.
This is used internally by the @invocation decorator logic. Do not use this directly.
"""
tags: Optional[list[str]] = Field(default_factory=None, description="The tags to display in the UI")
title: Optional[str] = Field(default=None, description="The display name of the node")
tags: Optional[list[str]] = Field(default_factory=None, description="The node's tags")
title: Optional[str] = Field(default=None, description="The node's display name")
category: Optional[str] = Field(default=None, description="The node's category")
version: Optional[str] = Field(
default=None, description='The node\'s version. Should be a valid semver string e.g. "1.0.0" or "3.8.13".'
)
class InvocationContext:
@@ -383,10 +421,11 @@ class InvocationContext:
class BaseInvocationOutput(BaseModel):
"""Base class for all invocation outputs"""
"""
Base class for all invocation outputs.
# All outputs must include a type name like this:
# type: Literal['your_output_name'] # noqa f821
All invocation outputs must use the `@invocation_output` decorator to provide their unique type.
"""
@classmethod
def get_all_subclasses_tuple(cls):
@@ -422,12 +461,12 @@ class MissingInputException(Exception):
class BaseInvocation(ABC, BaseModel):
"""A node to process inputs and produce outputs.
May use dependency injection in __init__ to receive providers.
"""
A node to process inputs and produce outputs.
May use dependency injection in __init__ to receive providers.
# All invocations must include a type name like this:
# type: Literal['your_output_name'] # noqa f821
All invocations must use the `@invocation` decorator to provide their unique type.
"""
@classmethod
def get_all_subclasses(cls):
@@ -466,6 +505,10 @@ class BaseInvocation(ABC, BaseModel):
schema["title"] = uiconfig.title
if uiconfig and hasattr(uiconfig, "tags"):
schema["tags"] = uiconfig.tags
if uiconfig and hasattr(uiconfig, "category"):
schema["category"] = uiconfig.category
if uiconfig and hasattr(uiconfig, "version"):
schema["version"] = uiconfig.version
if "required" not in schema or not isinstance(schema["required"], list):
schema["required"] = list()
schema["required"].extend(["type", "id"])
@@ -505,37 +548,124 @@ class BaseInvocation(ABC, BaseModel):
raise MissingInputException(self.__fields__["type"].default, field_name)
return self.invoke(context)
id: str = Field(description="The id of this node. Must be unique among all nodes.")
is_intermediate: bool = InputField(
default=False, description="Whether or not this node is an intermediate node.", input=Input.Direct
id: str = Field(
description="The id of this instance of an invocation. Must be unique among all instances of invocations."
)
is_intermediate: bool = InputField(
default=False, description="Whether or not this is an intermediate invocation.", ui_type=UIType.IsIntermediate
)
workflow: Optional[str] = InputField(
default=None,
description="The workflow to save with the image",
ui_type=UIType.WorkflowField,
)
@validator("workflow", pre=True)
def validate_workflow_is_json(cls, v):
if v is None:
return None
try:
json.loads(v)
except json.decoder.JSONDecodeError:
raise ValueError("Workflow must be valid JSON")
return v
UIConfig: ClassVar[Type[UIConfigBase]]
T = TypeVar("T", bound=BaseInvocation)
GenericBaseInvocation = TypeVar("GenericBaseInvocation", bound=BaseInvocation)
def title(title: str) -> Callable[[Type[T]], Type[T]]:
"""Adds a title to the invocation. Use this to override the default title generation, which is based on the class name."""
def invocation(
invocation_type: str,
title: Optional[str] = None,
tags: Optional[list[str]] = None,
category: Optional[str] = None,
version: Optional[str] = None,
) -> Callable[[Type[GenericBaseInvocation]], Type[GenericBaseInvocation]]:
"""
Adds metadata to an invocation.
def wrapper(cls: Type[T]) -> Type[T]:
:param str invocation_type: The type of the invocation. Must be unique among all invocations.
:param Optional[str] title: Adds a title to the invocation. Use if the auto-generated title isn't quite right. Defaults to None.
:param Optional[list[str]] tags: Adds tags to the invocation. Invocations may be searched for by their tags. Defaults to None.
:param Optional[str] category: Adds a category to the invocation. Used to group the invocations in the UI. Defaults to None.
"""
def wrapper(cls: Type[GenericBaseInvocation]) -> Type[GenericBaseInvocation]:
# Validate invocation types on creation of invocation classes
# TODO: ensure unique?
if re.compile(r"^\S+$").match(invocation_type) is None:
raise ValueError(f'"invocation_type" must consist of non-whitespace characters, got "{invocation_type}"')
# Add OpenAPI schema extras
uiconf_name = cls.__qualname__ + ".UIConfig"
if not hasattr(cls, "UIConfig") or cls.UIConfig.__qualname__ != uiconf_name:
cls.UIConfig = type(uiconf_name, (UIConfigBase,), dict())
cls.UIConfig.title = title
if title is not None:
cls.UIConfig.title = title
if tags is not None:
cls.UIConfig.tags = tags
if category is not None:
cls.UIConfig.category = category
if version is not None:
try:
semver.Version.parse(version)
except ValueError as e:
raise InvalidVersionError(f'Invalid version string for node "{invocation_type}": "{version}"') from e
cls.UIConfig.version = version
# Add the invocation type to the pydantic model of the invocation
invocation_type_annotation = Literal[invocation_type] # type: ignore
invocation_type_field = ModelField.infer(
name="type",
value=invocation_type,
annotation=invocation_type_annotation,
class_validators=None,
config=cls.__config__,
)
cls.__fields__.update({"type": invocation_type_field})
# to support 3.9, 3.10 and 3.11, as described in https://docs.python.org/3/howto/annotations.html
if annotations := cls.__dict__.get("__annotations__", None):
annotations.update({"type": invocation_type_annotation})
return cls
return wrapper
def tags(*tags: str) -> Callable[[Type[T]], Type[T]]:
"""Adds tags to the invocation. Use this to improve the streamline finding the invocation in the UI."""
GenericBaseInvocationOutput = TypeVar("GenericBaseInvocationOutput", bound=BaseInvocationOutput)
def invocation_output(
output_type: str,
) -> Callable[[Type[GenericBaseInvocationOutput]], Type[GenericBaseInvocationOutput]]:
"""
Adds metadata to an invocation output.
:param str output_type: The type of the invocation output. Must be unique among all invocation outputs.
"""
def wrapper(cls: Type[GenericBaseInvocationOutput]) -> Type[GenericBaseInvocationOutput]:
# Validate output types on creation of invocation output classes
# TODO: ensure unique?
if re.compile(r"^\S+$").match(output_type) is None:
raise ValueError(f'"output_type" must consist of non-whitespace characters, got "{output_type}"')
# Add the output type to the pydantic model of the invocation output
output_type_annotation = Literal[output_type] # type: ignore
output_type_field = ModelField.infer(
name="type",
value=output_type,
annotation=output_type_annotation,
class_validators=None,
config=cls.__config__,
)
cls.__fields__.update({"type": output_type_field})
# to support 3.9, 3.10 and 3.11, as described in https://docs.python.org/3/howto/annotations.html
if annotations := cls.__dict__.get("__annotations__", None):
annotations.update({"type": output_type_annotation})
def wrapper(cls: Type[T]) -> Type[T]:
uiconf_name = cls.__qualname__ + ".UIConfig"
if not hasattr(cls, "UIConfig") or cls.UIConfig.__qualname__ != uiconf_name:
cls.UIConfig = type(uiconf_name, (UIConfigBase,), dict())
cls.UIConfig.tags = list(tags)
return cls
return wrapper

35
invokeai/app/invocations/collections.py
View File

@@ -1,6 +1,5 @@
# Copyright (c) 2023 Kyle Schouviller (https://github.com/kyle0654) and the InvokeAI Team
from typing import Literal
import numpy as np
from pydantic import validator
@@ -8,17 +7,15 @@ from pydantic import validator
from invokeai.app.invocations.primitives import IntegerCollectionOutput
from invokeai.app.util.misc import SEED_MAX, get_random_seed
from .baseinvocation import BaseInvocation, InputField, InvocationContext, tags, title
from .baseinvocation import BaseInvocation, InputField, InvocationContext, invocation
@title("Integer Range")
@tags("collection", "integer", "range")
@invocation(
"range", title="Integer Range", tags=["collection", "integer", "range"], category="collections", version="1.0.0"
)
class RangeInvocation(BaseInvocation):
"""Creates a range of numbers from start to stop with step"""
type: Literal["range"] = "range"
# Inputs
start: int = InputField(default=0, description="The start of the range")
stop: int = InputField(default=10, description="The stop of the range")
step: int = InputField(default=1, description="The step of the range")
@@ -33,14 +30,16 @@ class RangeInvocation(BaseInvocation):
return IntegerCollectionOutput(collection=list(range(self.start, self.stop, self.step)))
@title("Integer Range of Size")
@tags("range", "integer", "size", "collection")
@invocation(
"range_of_size",
title="Integer Range of Size",
tags=["collection", "integer", "size", "range"],
category="collections",
version="1.0.0",
)
class RangeOfSizeInvocation(BaseInvocation):
"""Creates a range from start to start + size with step"""
type: Literal["range_of_size"] = "range_of_size"
# Inputs
start: int = InputField(default=0, description="The start of the range")
size: int = InputField(default=1, description="The number of values")
step: int = InputField(default=1, description="The step of the range")
@@ -49,14 +48,16 @@ class RangeOfSizeInvocation(BaseInvocation):
return IntegerCollectionOutput(collection=list(range(self.start, self.start + self.size, self.step)))
@title("Random Range")
@tags("range", "integer", "random", "collection")
@invocation(
"random_range",
title="Random Range",
tags=["range", "integer", "random", "collection"],
category="collections",
version="1.0.0",
)
class RandomRangeInvocation(BaseInvocation):
"""Creates a collection of random numbers"""
type: Literal["random_range"] = "random_range"
# Inputs
low: int = InputField(default=0, description="The inclusive low value")
high: int = InputField(default=np.iinfo(np.int32).max, description="The exclusive high value")
size: int = InputField(default=1, description="The number of values to generate")

79
invokeai/app/invocations/compel.py
View File

@@ -1,6 +1,6 @@
import re
from dataclasses import dataclass
from typing import List, Literal, Union
from typing import List, Union
import torch
from compel import Compel, ReturnedEmbeddingsType
@@ -26,8 +26,8 @@ from .baseinvocation import (
InvocationContext,
OutputField,
UIComponent,
tags,
title,
invocation,
invocation_output,
)
from .model import ClipField
@@ -44,13 +44,10 @@ class ConditioningFieldData:
# PerpNeg = "perp_neg"
@title("Compel Prompt")
@tags("prompt", "compel")
@invocation("compel", title="Prompt", tags=["prompt", "compel"], category="conditioning", version="1.0.0")
class CompelInvocation(BaseInvocation):
"""Parse prompt using compel package to conditioning."""
type: Literal["compel"] = "compel"
prompt: str = InputField(
default="",
description=FieldDescriptions.compel_prompt,
@@ -116,16 +113,15 @@ class CompelInvocation(BaseInvocation):
text_encoder=text_encoder,
textual_inversion_manager=ti_manager,
dtype_for_device_getter=torch_dtype,
truncate_long_prompts=True,
truncate_long_prompts=False,
)
conjunction = Compel.parse_prompt_string(self.prompt)
prompt: Union[FlattenedPrompt, Blend] = conjunction.prompts[0]
if context.services.configuration.log_tokenization:
log_tokenization_for_prompt_object(prompt, tokenizer)
log_tokenization_for_conjunction(conjunction, tokenizer)
c, options = compel.build_conditioning_tensor_for_prompt_object(prompt)
c, options = compel.build_conditioning_tensor_for_conjunction(conjunction)
ec = InvokeAIDiffuserComponent.ExtraConditioningInfo(
tokens_count_including_eos_bos=get_max_token_count(tokenizer, conjunction),
@@ -231,7 +227,7 @@ class SDXLPromptInvocationBase:
text_encoder=text_encoder,
textual_inversion_manager=ti_manager,
dtype_for_device_getter=torch_dtype,
truncate_long_prompts=True, # TODO:
truncate_long_prompts=False, # TODO:
returned_embeddings_type=ReturnedEmbeddingsType.PENULTIMATE_HIDDEN_STATES_NON_NORMALIZED, # TODO: clip skip
requires_pooled=get_pooled,
)
@@ -240,8 +236,7 @@ class SDXLPromptInvocationBase:
if context.services.configuration.log_tokenization:
# TODO: better logging for and syntax
for prompt_obj in conjunction.prompts:
log_tokenization_for_prompt_object(prompt_obj, tokenizer)
log_tokenization_for_conjunction(conjunction, tokenizer)
# TODO: ask for optimizations? to not run text_encoder twice
c, options = compel.build_conditioning_tensor_for_conjunction(conjunction)
@@ -267,13 +262,16 @@ class SDXLPromptInvocationBase:
return c, c_pooled, ec
@title("SDXL Compel Prompt")
@tags("sdxl", "compel", "prompt")
@invocation(
"sdxl_compel_prompt",
title="SDXL Prompt",
tags=["sdxl", "compel", "prompt"],
category="conditioning",
version="1.0.0",
)
class SDXLCompelPromptInvocation(BaseInvocation, SDXLPromptInvocationBase):
"""Parse prompt using compel package to conditioning."""
type: Literal["sdxl_compel_prompt"] = "sdxl_compel_prompt"
prompt: str = InputField(default="", description=FieldDescriptions.compel_prompt, ui_component=UIComponent.Textarea)
style: str = InputField(default="", description=FieldDescriptions.compel_prompt, ui_component=UIComponent.Textarea)
original_width: int = InputField(default=1024, description="")
@@ -282,8 +280,8 @@ class SDXLCompelPromptInvocation(BaseInvocation, SDXLPromptInvocationBase):
crop_left: int = InputField(default=0, description="")
target_width: int = InputField(default=1024, description="")
target_height: int = InputField(default=1024, description="")
clip: ClipField = InputField(description=FieldDescriptions.clip, input=Input.Connection)
clip2: ClipField = InputField(description=FieldDescriptions.clip, input=Input.Connection)
clip: ClipField = InputField(description=FieldDescriptions.clip, input=Input.Connection, title="CLIP 1")
clip2: ClipField = InputField(description=FieldDescriptions.clip, input=Input.Connection, title="CLIP 2")
@torch.no_grad()
def invoke(self, context: InvocationContext) -> ConditioningOutput:
@@ -305,6 +303,29 @@ class SDXLCompelPromptInvocation(BaseInvocation, SDXLPromptInvocationBase):
add_time_ids = torch.tensor([original_size + crop_coords + target_size])
# [1, 77, 768], [1, 154, 1280]
if c1.shape[1] < c2.shape[1]:
c1 = torch.cat(
[
c1,
torch.zeros(
(c1.shape[0], c2.shape[1] - c1.shape[1], c1.shape[2]), device=c1.device, dtype=c1.dtype
),
],
dim=1,
)
elif c1.shape[1] > c2.shape[1]:
c2 = torch.cat(
[
c2,
torch.zeros(
(c2.shape[0], c1.shape[1] - c2.shape[1], c2.shape[2]), device=c2.device, dtype=c2.dtype
),
],
dim=1,
)
conditioning_data = ConditioningFieldData(
conditionings=[
SDXLConditioningInfo(
@@ -326,13 +347,16 @@ class SDXLCompelPromptInvocation(BaseInvocation, SDXLPromptInvocationBase):
)
@title("SDXL Refiner Compel Prompt")
@tags("sdxl", "compel", "prompt")
@invocation(
"sdxl_refiner_compel_prompt",
title="SDXL Refiner Prompt",
tags=["sdxl", "compel", "prompt"],
category="conditioning",
version="1.0.0",
)
class SDXLRefinerCompelPromptInvocation(BaseInvocation, SDXLPromptInvocationBase):
"""Parse prompt using compel package to conditioning."""
type: Literal["sdxl_refiner_compel_prompt"] = "sdxl_refiner_compel_prompt"
style: str = InputField(
default="", description=FieldDescriptions.compel_prompt, ui_component=UIComponent.Textarea
) # TODO: ?
@@ -374,20 +398,17 @@ class SDXLRefinerCompelPromptInvocation(BaseInvocation, SDXLPromptInvocationBase
)
@invocation_output("clip_skip_output")
class ClipSkipInvocationOutput(BaseInvocationOutput):
"""Clip skip node output"""
type: Literal["clip_skip_output"] = "clip_skip_output"
clip: ClipField = OutputField(default=None, description=FieldDescriptions.clip, title="CLIP")
@title("CLIP Skip")
@tags("clipskip", "clip", "skip")
@invocation("clip_skip", title="CLIP Skip", tags=["clipskip", "clip", "skip"], category="conditioning", version="1.0.0")
class ClipSkipInvocation(BaseInvocation):
"""Skip layers in clip text_encoder model."""
type: Literal["clip_skip"] = "clip_skip"
clip: ClipField = InputField(description=FieldDescriptions.clip, input=Input.Connection, title="CLIP")
skipped_layers: int = InputField(default=0, description=FieldDescriptions.skipped_layers)

199
invokeai/app/invocations/controlnet_image_processors.py
View File

@@ -40,8 +40,8 @@ from .baseinvocation import (
InvocationContext,
OutputField,
UIType,
tags,
title,
invocation,
invocation_output,
)
@@ -87,27 +87,20 @@ class ControlField(BaseModel):
return v
@invocation_output("control_output")
class ControlOutput(BaseInvocationOutput):
"""node output for ControlNet info"""
type: Literal["control_output"] = "control_output"
# Outputs
control: ControlField = OutputField(description=FieldDescriptions.control)
@title("ControlNet")
@tags("controlnet")
@invocation("controlnet", title="ControlNet", tags=["controlnet"], category="controlnet", version="1.0.0")
class ControlNetInvocation(BaseInvocation):
"""Collects ControlNet info to pass to other nodes"""
type: Literal["controlnet"] = "controlnet"
# Inputs
image: ImageField = InputField(description="The control image")
control_model: ControlNetModelField = InputField(
default="lllyasviel/sd-controlnet-canny", description=FieldDescriptions.controlnet_model, input=Input.Direct
)
control_model: ControlNetModelField = InputField(description=FieldDescriptions.controlnet_model, input=Input.Direct)
control_weight: Union[float, List[float]] = InputField(
default=1.0, description="The weight given to the ControlNet", ui_type=UIType.Float
)
@@ -134,12 +127,12 @@ class ControlNetInvocation(BaseInvocation):
)
@invocation(
"image_processor", title="Base Image Processor", tags=["controlnet"], category="controlnet", version="1.0.0"
)
class ImageProcessorInvocation(BaseInvocation):
"""Base class for invocations that preprocess images for ControlNet"""
type: Literal["image_processor"] = "image_processor"
# Inputs
image: ImageField = InputField(description="The image to process")
def run_processor(self, image):
@@ -151,11 +144,6 @@ class ImageProcessorInvocation(BaseInvocation):
# image type should be PIL.PngImagePlugin.PngImageFile ?
processed_image = self.run_processor(raw_image)
# FIXME: what happened to image metadata?
# metadata = context.services.metadata.build_metadata(
# session_id=context.graph_execution_state_id, node=self
# )
# currently can't see processed image in node UI without a showImage node,
# so for now setting image_type to RESULT instead of INTERMEDIATE so will get saved in gallery
image_dto = context.services.images.create(
@@ -165,6 +153,7 @@ class ImageProcessorInvocation(BaseInvocation):
session_id=context.graph_execution_state_id,
node_id=self.id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
"""Builds an ImageOutput and its ImageField"""
@@ -179,14 +168,16 @@ class ImageProcessorInvocation(BaseInvocation):
)
@title("Canny Processor")
@tags("controlnet", "canny")
@invocation(
"canny_image_processor",
title="Canny Processor",
tags=["controlnet", "canny"],
category="controlnet",
version="1.0.0",
)
class CannyImageProcessorInvocation(ImageProcessorInvocation):
"""Canny edge detection for ControlNet"""
type: Literal["canny_image_processor"] = "canny_image_processor"
# Input
low_threshold: int = InputField(
default=100, ge=0, le=255, description="The low threshold of the Canny pixel gradient (0-255)"
)
@@ -200,14 +191,16 @@ class CannyImageProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("HED (softedge) Processor")
@tags("controlnet", "hed", "softedge")
@invocation(
"hed_image_processor",
title="HED (softedge) Processor",
tags=["controlnet", "hed", "softedge"],
category="controlnet",
version="1.0.0",
)
class HedImageProcessorInvocation(ImageProcessorInvocation):
"""Applies HED edge detection to image"""
type: Literal["hed_image_processor"] = "hed_image_processor"
# Inputs
detect_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.detect_res)
image_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.image_res)
# safe not supported in controlnet_aux v0.0.3
@@ -227,14 +220,16 @@ class HedImageProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("Lineart Processor")
@tags("controlnet", "lineart")
@invocation(
"lineart_image_processor",
title="Lineart Processor",
tags=["controlnet", "lineart"],
category="controlnet",
version="1.0.0",
)
class LineartImageProcessorInvocation(ImageProcessorInvocation):
"""Applies line art processing to image"""
type: Literal["lineart_image_processor"] = "lineart_image_processor"
# Inputs
detect_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.detect_res)
image_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.image_res)
coarse: bool = InputField(default=False, description="Whether to use coarse mode")
@@ -247,14 +242,16 @@ class LineartImageProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("Lineart Anime Processor")
@tags("controlnet", "lineart", "anime")
@invocation(
"lineart_anime_image_processor",
title="Lineart Anime Processor",
tags=["controlnet", "lineart", "anime"],
category="controlnet",
version="1.0.0",
)
class LineartAnimeImageProcessorInvocation(ImageProcessorInvocation):
"""Applies line art anime processing to image"""
type: Literal["lineart_anime_image_processor"] = "lineart_anime_image_processor"
# Inputs
detect_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.detect_res)
image_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.image_res)
@@ -268,14 +265,16 @@ class LineartAnimeImageProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("Openpose Processor")
@tags("controlnet", "openpose", "pose")
@invocation(
"openpose_image_processor",
title="Openpose Processor",
tags=["controlnet", "openpose", "pose"],
category="controlnet",
version="1.0.0",
)
class OpenposeImageProcessorInvocation(ImageProcessorInvocation):
"""Applies Openpose processing to image"""
type: Literal["openpose_image_processor"] = "openpose_image_processor"
# Inputs
hand_and_face: bool = InputField(default=False, description="Whether to use hands and face mode")
detect_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.detect_res)
image_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.image_res)
@@ -291,14 +290,16 @@ class OpenposeImageProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("Midas (Depth) Processor")
@tags("controlnet", "midas", "depth")
@invocation(
"midas_depth_image_processor",
title="Midas Depth Processor",
tags=["controlnet", "midas"],
category="controlnet",
version="1.0.0",
)
class MidasDepthImageProcessorInvocation(ImageProcessorInvocation):
"""Applies Midas depth processing to image"""
type: Literal["midas_depth_image_processor"] = "midas_depth_image_processor"
# Inputs
a_mult: float = InputField(default=2.0, ge=0, description="Midas parameter `a_mult` (a = a_mult * PI)")
bg_th: float = InputField(default=0.1, ge=0, description="Midas parameter `bg_th`")
# depth_and_normal not supported in controlnet_aux v0.0.3
@@ -316,14 +317,16 @@ class MidasDepthImageProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("Normal BAE Processor")
@tags("controlnet", "normal", "bae")
@invocation(
"normalbae_image_processor",
title="Normal BAE Processor",
tags=["controlnet"],
category="controlnet",
version="1.0.0",
)
class NormalbaeImageProcessorInvocation(ImageProcessorInvocation):
"""Applies NormalBae processing to image"""
type: Literal["normalbae_image_processor"] = "normalbae_image_processor"
# Inputs
detect_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.detect_res)
image_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.image_res)
@@ -335,14 +338,12 @@ class NormalbaeImageProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("MLSD Processor")
@tags("controlnet", "mlsd")
@invocation(
"mlsd_image_processor", title="MLSD Processor", tags=["controlnet", "mlsd"], category="controlnet", version="1.0.0"
)
class MlsdImageProcessorInvocation(ImageProcessorInvocation):
"""Applies MLSD processing to image"""
type: Literal["mlsd_image_processor"] = "mlsd_image_processor"
# Inputs
detect_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.detect_res)
image_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.image_res)
thr_v: float = InputField(default=0.1, ge=0, description="MLSD parameter `thr_v`")
@@ -360,14 +361,12 @@ class MlsdImageProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("PIDI Processor")
@tags("controlnet", "pidi")
@invocation(
"pidi_image_processor", title="PIDI Processor", tags=["controlnet", "pidi"], category="controlnet", version="1.0.0"
)
class PidiImageProcessorInvocation(ImageProcessorInvocation):
"""Applies PIDI processing to image"""
type: Literal["pidi_image_processor"] = "pidi_image_processor"
# Inputs
detect_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.detect_res)
image_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.image_res)
safe: bool = InputField(default=False, description=FieldDescriptions.safe_mode)
@@ -385,14 +384,16 @@ class PidiImageProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("Content Shuffle Processor")
@tags("controlnet", "contentshuffle")
@invocation(
"content_shuffle_image_processor",
title="Content Shuffle Processor",
tags=["controlnet", "contentshuffle"],
category="controlnet",
version="1.0.0",
)
class ContentShuffleImageProcessorInvocation(ImageProcessorInvocation):
"""Applies content shuffle processing to image"""
type: Literal["content_shuffle_image_processor"] = "content_shuffle_image_processor"
# Inputs
detect_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.detect_res)
image_resolution: int = InputField(default=512, ge=0, description=FieldDescriptions.image_res)
h: Optional[int] = InputField(default=512, ge=0, description="Content shuffle `h` parameter")
@@ -413,27 +414,32 @@ class ContentShuffleImageProcessorInvocation(ImageProcessorInvocation):
# should work with controlnet_aux >= 0.0.4 and timm <= 0.6.13
@title("Zoe (Depth) Processor")
@tags("controlnet", "zoe", "depth")
@invocation(
"zoe_depth_image_processor",
title="Zoe (Depth) Processor",
tags=["controlnet", "zoe", "depth"],
category="controlnet",
version="1.0.0",
)
class ZoeDepthImageProcessorInvocation(ImageProcessorInvocation):
"""Applies Zoe depth processing to image"""
type: Literal["zoe_depth_image_processor"] = "zoe_depth_image_processor"
def run_processor(self, image):
zoe_depth_processor = ZoeDetector.from_pretrained("lllyasviel/Annotators")
processed_image = zoe_depth_processor(image)
return processed_image
@title("Mediapipe Face Processor")
@tags("controlnet", "mediapipe", "face")
@invocation(
"mediapipe_face_processor",
title="Mediapipe Face Processor",
tags=["controlnet", "mediapipe", "face"],
category="controlnet",
version="1.0.0",
)
class MediapipeFaceProcessorInvocation(ImageProcessorInvocation):
"""Applies mediapipe face processing to image"""
type: Literal["mediapipe_face_processor"] = "mediapipe_face_processor"
# Inputs
max_faces: int = InputField(default=1, ge=1, description="Maximum number of faces to detect")
min_confidence: float = InputField(default=0.5, ge=0, le=1, description="Minimum confidence for face detection")
@@ -447,14 +453,16 @@ class MediapipeFaceProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("Leres (Depth) Processor")
@tags("controlnet", "leres", "depth")
@invocation(
"leres_image_processor",
title="Leres (Depth) Processor",
tags=["controlnet", "leres", "depth"],
category="controlnet",
version="1.0.0",
)
class LeresImageProcessorInvocation(ImageProcessorInvocation):
"""Applies leres processing to image"""
type: Literal["leres_image_processor"] = "leres_image_processor"
# Inputs
thr_a: float = InputField(default=0, description="Leres parameter `thr_a`")
thr_b: float = InputField(default=0, description="Leres parameter `thr_b`")
boost: bool = InputField(default=False, description="Whether to use boost mode")
@@ -474,14 +482,16 @@ class LeresImageProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("Tile Resample Processor")
@tags("controlnet", "tile")
@invocation(
"tile_image_processor",
title="Tile Resample Processor",
tags=["controlnet", "tile"],
category="controlnet",
version="1.0.0",
)
class TileResamplerProcessorInvocation(ImageProcessorInvocation):
"""Tile resampler processor"""
type: Literal["tile_image_processor"] = "tile_image_processor"
# Inputs
# res: int = InputField(default=512, ge=0, le=1024, description="The pixel resolution for each tile")
down_sampling_rate: float = InputField(default=1.0, ge=1.0, le=8.0, description="Down sampling rate")
@@ -512,13 +522,16 @@ class TileResamplerProcessorInvocation(ImageProcessorInvocation):
return processed_image
@title("Segment Anything Processor")
@tags("controlnet", "segmentanything")
@invocation(
"segment_anything_processor",
title="Segment Anything Processor",
tags=["controlnet", "segmentanything"],
category="controlnet",
version="1.0.0",
)
class SegmentAnythingProcessorInvocation(ImageProcessorInvocation):
"""Applies segment anything processing to image"""
type: Literal["segment_anything_processor"] = "segment_anything_processor"
def run_processor(self, image):
# segment_anything_processor = SamDetector.from_pretrained("ybelkada/segment-anything", subfolder="checkpoints")
segment_anything_processor = SamDetectorReproducibleColors.from_pretrained(

10
invokeai/app/invocations/cv.py
View File

@@ -1,6 +1,5 @@
# Copyright (c) 2022 Kyle Schouviller (https://github.com/kyle0654)
from typing import Literal
import cv2 as cv
import numpy
@@ -8,17 +7,13 @@ from PIL import Image, ImageOps
from invokeai.app.invocations.primitives import ImageField, ImageOutput
from invokeai.app.models.image import ImageCategory, ResourceOrigin
from .baseinvocation import BaseInvocation, InputField, InvocationContext, tags, title
from .baseinvocation import BaseInvocation, InputField, InvocationContext, invocation
@title("OpenCV Inpaint")
@tags("opencv", "inpaint")
@invocation("cv_inpaint", title="OpenCV Inpaint", tags=["opencv", "inpaint"], category="inpaint", version="1.0.0")
class CvInpaintInvocation(BaseInvocation):
"""Simple inpaint using opencv."""
type: Literal["cv_inpaint"] = "cv_inpaint"
# Inputs
image: ImageField = InputField(description="The image to inpaint")
mask: ImageField = InputField(description="The mask to use when inpainting")
@@ -45,6 +40,7 @@ class CvInpaintInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(

185
invokeai/app/invocations/image.py
View File

@@ -13,18 +13,13 @@ from invokeai.backend.image_util.invisible_watermark import InvisibleWatermark
from invokeai.backend.image_util.safety_checker import SafetyChecker
from ..models.image import ImageCategory, ResourceOrigin
from .baseinvocation import BaseInvocation, FieldDescriptions, InputField, InvocationContext, tags, title
from .baseinvocation import BaseInvocation, FieldDescriptions, InputField, InvocationContext, invocation
@title("Show Image")
@tags("image")
@invocation("show_image", title="Show Image", tags=["image"], category="image", version="1.0.0")
class ShowImageInvocation(BaseInvocation):
"""Displays a provided image, and passes it forward in the pipeline."""
"""Displays a provided image using the OS image viewer, and passes it forward in the pipeline."""
# Metadata
type: Literal["show_image"] = "show_image"
# Inputs
image: ImageField = InputField(description="The image to show")
def invoke(self, context: InvocationContext) -> ImageOutput:
@@ -41,15 +36,10 @@ class ShowImageInvocation(BaseInvocation):
)
@title("Blank Image")
@tags("image")
@invocation("blank_image", title="Blank Image", tags=["image"], category="image", version="1.0.0")
class BlankImageInvocation(BaseInvocation):
"""Creates a blank image and forwards it to the pipeline"""
# Metadata
type: Literal["blank_image"] = "blank_image"
# Inputs
width: int = InputField(default=512, description="The width of the image")
height: int = InputField(default=512, description="The height of the image")
mode: Literal["RGB", "RGBA"] = InputField(default="RGB", description="The mode of the image")
@@ -65,6 +55,7 @@ class BlankImageInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -74,15 +65,10 @@ class BlankImageInvocation(BaseInvocation):
)
@title("Crop Image")
@tags("image", "crop")
@invocation("img_crop", title="Crop Image", tags=["image", "crop"], category="image", version="1.0.0")
class ImageCropInvocation(BaseInvocation):
"""Crops an image to a specified box. The box can be outside of the image."""
# Metadata
type: Literal["img_crop"] = "img_crop"
# Inputs
image: ImageField = InputField(description="The image to crop")
x: int = InputField(default=0, description="The left x coordinate of the crop rectangle")
y: int = InputField(default=0, description="The top y coordinate of the crop rectangle")
@@ -102,6 +88,7 @@ class ImageCropInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -111,15 +98,10 @@ class ImageCropInvocation(BaseInvocation):
)
@title("Paste Image")
@tags("image", "paste")
@invocation("img_paste", title="Paste Image", tags=["image", "paste"], category="image", version="1.0.0")
class ImagePasteInvocation(BaseInvocation):
"""Pastes an image into another image."""
# Metadata
type: Literal["img_paste"] = "img_paste"
# Inputs
base_image: ImageField = InputField(description="The base image")
image: ImageField = InputField(description="The image to paste")
mask: Optional[ImageField] = InputField(
@@ -154,6 +136,7 @@ class ImagePasteInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -163,15 +146,10 @@ class ImagePasteInvocation(BaseInvocation):
)
@title("Mask from Alpha")
@tags("image", "mask")
@invocation("tomask", title="Mask from Alpha", tags=["image", "mask"], category="image", version="1.0.0")
class MaskFromAlphaInvocation(BaseInvocation):
"""Extracts the alpha channel of an image as a mask."""
# Metadata
type: Literal["tomask"] = "tomask"
# Inputs
image: ImageField = InputField(description="The image to create the mask from")
invert: bool = InputField(default=False, description="Whether or not to invert the mask")
@@ -189,6 +167,7 @@ class MaskFromAlphaInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -198,15 +177,10 @@ class MaskFromAlphaInvocation(BaseInvocation):
)
@title("Multiply Images")
@tags("image", "multiply")
@invocation("img_mul", title="Multiply Images", tags=["image", "multiply"], category="image", version="1.0.0")
class ImageMultiplyInvocation(BaseInvocation):
"""Multiplies two images together using `PIL.ImageChops.multiply()`."""
# Metadata
type: Literal["img_mul"] = "img_mul"
# Inputs
image1: ImageField = InputField(description="The first image to multiply")
image2: ImageField = InputField(description="The second image to multiply")
@@ -223,6 +197,7 @@ class ImageMultiplyInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -235,15 +210,10 @@ class ImageMultiplyInvocation(BaseInvocation):
IMAGE_CHANNELS = Literal["A", "R", "G", "B"]
@title("Extract Image Channel")
@tags("image", "channel")
@invocation("img_chan", title="Extract Image Channel", tags=["image", "channel"], category="image", version="1.0.0")
class ImageChannelInvocation(BaseInvocation):
"""Gets a channel from an image."""
# Metadata
type: Literal["img_chan"] = "img_chan"
# Inputs
image: ImageField = InputField(description="The image to get the channel from")
channel: IMAGE_CHANNELS = InputField(default="A", description="The channel to get")
@@ -259,6 +229,7 @@ class ImageChannelInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -271,15 +242,10 @@ class ImageChannelInvocation(BaseInvocation):
IMAGE_MODES = Literal["L", "RGB", "RGBA", "CMYK", "YCbCr", "LAB", "HSV", "I", "F"]
@title("Convert Image Mode")
@tags("image", "convert")
@invocation("img_conv", title="Convert Image Mode", tags=["image", "convert"], category="image", version="1.0.0")
class ImageConvertInvocation(BaseInvocation):
"""Converts an image to a different mode."""
# Metadata
type: Literal["img_conv"] = "img_conv"
# Inputs
image: ImageField = InputField(description="The image to convert")
mode: IMAGE_MODES = InputField(default="L", description="The mode to convert to")
@@ -295,6 +261,7 @@ class ImageConvertInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -304,15 +271,10 @@ class ImageConvertInvocation(BaseInvocation):
)
@title("Blur Image")
@tags("image", "blur")
@invocation("img_blur", title="Blur Image", tags=["image", "blur"], category="image", version="1.0.0")
class ImageBlurInvocation(BaseInvocation):
"""Blurs an image"""
# Metadata
type: Literal["img_blur"] = "img_blur"
# Inputs
image: ImageField = InputField(description="The image to blur")
radius: float = InputField(default=8.0, ge=0, description="The blur radius")
# Metadata
@@ -333,6 +295,7 @@ class ImageBlurInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -362,19 +325,17 @@ PIL_RESAMPLING_MAP = {
}
@title("Resize Image")
@tags("image", "resize")
@invocation("img_resize", title="Resize Image", tags=["image", "resize"], category="image", version="1.0.0")
class ImageResizeInvocation(BaseInvocation):
"""Resizes an image to specific dimensions"""
# Metadata
type: Literal["img_resize"] = "img_resize"
# Inputs
image: ImageField = InputField(description="The image to resize")
width: int = InputField(default=512, ge=64, multiple_of=8, description="The width to resize to (px)")
height: int = InputField(default=512, ge=64, multiple_of=8, description="The height to resize to (px)")
resample_mode: PIL_RESAMPLING_MODES = InputField(default="bicubic", description="The resampling mode")
metadata: Optional[CoreMetadata] = InputField(
default=None, description=FieldDescriptions.core_metadata, ui_hidden=True
)
def invoke(self, context: InvocationContext) -> ImageOutput:
image = context.services.images.get_pil_image(self.image.image_name)
@@ -393,6 +354,8 @@ class ImageResizeInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
metadata=self.metadata.dict() if self.metadata else None,
workflow=self.workflow,
)
return ImageOutput(
@@ -402,15 +365,10 @@ class ImageResizeInvocation(BaseInvocation):
)
@title("Scale Image")
@tags("image", "scale")
@invocation("img_scale", title="Scale Image", tags=["image", "scale"], category="image", version="1.0.0")
class ImageScaleInvocation(BaseInvocation):
"""Scales an image by a factor"""
# Metadata
type: Literal["img_scale"] = "img_scale"
# Inputs
image: ImageField = InputField(description="The image to scale")
scale_factor: float = InputField(
default=2.0,
@@ -438,6 +396,7 @@ class ImageScaleInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -447,15 +406,10 @@ class ImageScaleInvocation(BaseInvocation):
)
@title("Lerp Image")
@tags("image", "lerp")
@invocation("img_lerp", title="Lerp Image", tags=["image", "lerp"], category="image", version="1.0.0")
class ImageLerpInvocation(BaseInvocation):
"""Linear interpolation of all pixels of an image"""
# Metadata
type: Literal["img_lerp"] = "img_lerp"
# Inputs
image: ImageField = InputField(description="The image to lerp")
min: int = InputField(default=0, ge=0, le=255, description="The minimum output value")
max: int = InputField(default=255, ge=0, le=255, description="The maximum output value")
@@ -475,6 +429,7 @@ class ImageLerpInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -484,15 +439,10 @@ class ImageLerpInvocation(BaseInvocation):
)
@title("Inverse Lerp Image")
@tags("image", "ilerp")
@invocation("img_ilerp", title="Inverse Lerp Image", tags=["image", "ilerp"], category="image", version="1.0.0")
class ImageInverseLerpInvocation(BaseInvocation):
"""Inverse linear interpolation of all pixels of an image"""
# Metadata
type: Literal["img_ilerp"] = "img_ilerp"
# Inputs
image: ImageField = InputField(description="The image to lerp")
min: int = InputField(default=0, ge=0, le=255, description="The minimum input value")
max: int = InputField(default=255, ge=0, le=255, description="The maximum input value")
@@ -512,6 +462,7 @@ class ImageInverseLerpInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -521,15 +472,10 @@ class ImageInverseLerpInvocation(BaseInvocation):
)
@title("Blur NSFW Image")
@tags("image", "nsfw")
@invocation("img_nsfw", title="Blur NSFW Image", tags=["image", "nsfw"], category="image", version="1.0.0")
class ImageNSFWBlurInvocation(BaseInvocation):
"""Add blur to NSFW-flagged images"""
# Metadata
type: Literal["img_nsfw"] = "img_nsfw"
# Inputs
image: ImageField = InputField(description="The image to check")
metadata: Optional[CoreMetadata] = InputField(
default=None, description=FieldDescriptions.core_metadata, ui_hidden=True
@@ -555,6 +501,7 @@ class ImageNSFWBlurInvocation(BaseInvocation):
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
metadata=self.metadata.dict() if self.metadata else None,
workflow=self.workflow,
)
return ImageOutput(
@@ -570,15 +517,12 @@ class ImageNSFWBlurInvocation(BaseInvocation):
return caution.resize((caution.width // 2, caution.height // 2))
@title("Add Invisible Watermark")
@tags("image", "watermark")
@invocation(
"img_watermark", title="Add Invisible Watermark", tags=["image", "watermark"], category="image", version="1.0.0"
)
class ImageWatermarkInvocation(BaseInvocation):
"""Add an invisible watermark to an image"""
# Metadata
type: Literal["img_watermark"] = "img_watermark"
# Inputs
image: ImageField = InputField(description="The image to check")
text: str = InputField(default="InvokeAI", description="Watermark text")
metadata: Optional[CoreMetadata] = InputField(
@@ -596,6 +540,7 @@ class ImageWatermarkInvocation(BaseInvocation):
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
metadata=self.metadata.dict() if self.metadata else None,
workflow=self.workflow,
)
return ImageOutput(
@@ -605,14 +550,10 @@ class ImageWatermarkInvocation(BaseInvocation):
)
@title("Mask Edge")
@tags("image", "mask", "inpaint")
@invocation("mask_edge", title="Mask Edge", tags=["image", "mask", "inpaint"], category="image", version="1.0.0")
class MaskEdgeInvocation(BaseInvocation):
"""Applies an edge mask to an image"""
type: Literal["mask_edge"] = "mask_edge"
# Inputs
image: ImageField = InputField(description="The image to apply the mask to")
edge_size: int = InputField(description="The size of the edge")
edge_blur: int = InputField(description="The amount of blur on the edge")
@@ -644,6 +585,7 @@ class MaskEdgeInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -653,14 +595,12 @@ class MaskEdgeInvocation(BaseInvocation):
)
@title("Combine Mask")
@tags("image", "mask", "multiply")
@invocation(
"mask_combine", title="Combine Masks", tags=["image", "mask", "multiply"], category="image", version="1.0.0"
)
class MaskCombineInvocation(BaseInvocation):
"""Combine two masks together by multiplying them using `PIL.ImageChops.multiply()`."""
type: Literal["mask_combine"] = "mask_combine"
# Inputs
mask1: ImageField = InputField(description="The first mask to combine")
mask2: ImageField = InputField(description="The second image to combine")
@@ -677,6 +617,7 @@ class MaskCombineInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -686,17 +627,13 @@ class MaskCombineInvocation(BaseInvocation):
)
@title("Color Correct")
@tags("image", "color")
@invocation("color_correct", title="Color Correct", tags=["image", "color"], category="image", version="1.0.0")
class ColorCorrectInvocation(BaseInvocation):
"""
Shifts the colors of a target image to match the reference image, optionally
using a mask to only color-correct certain regions of the target image.
"""
type: Literal["color_correct"] = "color_correct"
# Inputs
image: ImageField = InputField(description="The image to color-correct")
reference: ImageField = InputField(description="Reference image for color-correction")
mask: Optional[ImageField] = InputField(default=None, description="Mask to use when applying color-correction")
@@ -785,6 +722,7 @@ class ColorCorrectInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -794,14 +732,10 @@ class ColorCorrectInvocation(BaseInvocation):
)
@title("Image Hue Adjustment")
@tags("image", "hue", "hsl")
@invocation("img_hue_adjust", title="Adjust Image Hue", tags=["image", "hue"], category="image", version="1.0.0")
class ImageHueAdjustmentInvocation(BaseInvocation):
"""Adjusts the Hue of an image."""
type: Literal["img_hue_adjust"] = "img_hue_adjust"
# Inputs
image: ImageField = InputField(description="The image to adjust")
hue: int = InputField(default=0, description="The degrees by which to rotate the hue, 0-360")
@@ -827,6 +761,7 @@ class ImageHueAdjustmentInvocation(BaseInvocation):
node_id=self.id,
is_intermediate=self.is_intermediate,
session_id=context.graph_execution_state_id,
workflow=self.workflow,
)
return ImageOutput(
@@ -838,14 +773,16 @@ class ImageHueAdjustmentInvocation(BaseInvocation):
)
@title("Image Luminosity Adjustment")
@tags("image", "luminosity", "hsl")
@invocation(
"img_luminosity_adjust",
title="Adjust Image Luminosity",
tags=["image", "luminosity", "hsl"],
category="image",
version="1.0.0",
)
class ImageLuminosityAdjustmentInvocation(BaseInvocation):
"""Adjusts the Luminosity (Value) of an image."""
type: Literal["img_luminosity_adjust"] = "img_luminosity_adjust"
# Inputs
image: ImageField = InputField(description="The image to adjust")
luminosity: float = InputField(
default=1.0, ge=0, le=1, description="The factor by which to adjust the luminosity (value)"
@@ -877,6 +814,7 @@ class ImageLuminosityAdjustmentInvocation(BaseInvocation):
node_id=self.id,
is_intermediate=self.is_intermediate,
session_id=context.graph_execution_state_id,
workflow=self.workflow,
)
return ImageOutput(
@@ -888,14 +826,16 @@ class ImageLuminosityAdjustmentInvocation(BaseInvocation):
)
@title("Image Saturation Adjustment")
@tags("image", "saturation", "hsl")
@invocation(
"img_saturation_adjust",
title="Adjust Image Saturation",
tags=["image", "saturation", "hsl"],
category="image",
version="1.0.0",
)
class ImageSaturationAdjustmentInvocation(BaseInvocation):
"""Adjusts the Saturation of an image."""
type: Literal["img_saturation_adjust"] = "img_saturation_adjust"
# Inputs
image: ImageField = InputField(description="The image to adjust")
saturation: float = InputField(default=1.0, ge=0, le=1, description="The factor by which to adjust the saturation")
@@ -925,6 +865,7 @@ class ImageSaturationAdjustmentInvocation(BaseInvocation):
node_id=self.id,
is_intermediate=self.is_intermediate,
session_id=context.graph_execution_state_id,
workflow=self.workflow,
)
return ImageOutput(

31
invokeai/app/invocations/infill.py
View File

@@ -12,7 +12,7 @@ from invokeai.backend.image_util.lama import LaMA
from invokeai.backend.image_util.patchmatch import PatchMatch
from ..models.image import ImageCategory, ResourceOrigin
from .baseinvocation import BaseInvocation, InputField, InvocationContext, tags, title
from .baseinvocation import BaseInvocation, InputField, InvocationContext, invocation
def infill_methods() -> list[str]:
@@ -116,14 +116,10 @@ def tile_fill_missing(im: Image.Image, tile_size: int = 16, seed: Optional[int]
return si
@title("Solid Color Infill")
@tags("image", "inpaint")
@invocation("infill_rgba", title="Solid Color Infill", tags=["image", "inpaint"], category="inpaint", version="1.0.0")
class InfillColorInvocation(BaseInvocation):
"""Infills transparent areas of an image with a solid color"""
type: Literal["infill_rgba"] = "infill_rgba"
# Inputs
image: ImageField = InputField(description="The image to infill")
color: ColorField = InputField(
default=ColorField(r=127, g=127, b=127, a=255),
@@ -145,6 +141,7 @@ class InfillColorInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -154,14 +151,10 @@ class InfillColorInvocation(BaseInvocation):
)
@title("Tile Infill")
@tags("image", "inpaint")
@invocation("infill_tile", title="Tile Infill", tags=["image", "inpaint"], category="inpaint", version="1.0.0")
class InfillTileInvocation(BaseInvocation):
"""Infills transparent areas of an image with tiles of the image"""
type: Literal["infill_tile"] = "infill_tile"
# Input
image: ImageField = InputField(description="The image to infill")
tile_size: int = InputField(default=32, ge=1, description="The tile size (px)")
seed: int = InputField(
@@ -184,6 +177,7 @@ class InfillTileInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -193,14 +187,12 @@ class InfillTileInvocation(BaseInvocation):
)
@title("PatchMatch Infill")
@tags("image", "inpaint")
@invocation(
"infill_patchmatch", title="PatchMatch Infill", tags=["image", "inpaint"], category="inpaint", version="1.0.0"
)
class InfillPatchMatchInvocation(BaseInvocation):
"""Infills transparent areas of an image using the PatchMatch algorithm"""
type: Literal["infill_patchmatch"] = "infill_patchmatch"
# Inputs
image: ImageField = InputField(description="The image to infill")
def invoke(self, context: InvocationContext) -> ImageOutput:
@@ -218,6 +210,7 @@ class InfillPatchMatchInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(
@@ -227,14 +220,10 @@ class InfillPatchMatchInvocation(BaseInvocation):
)
@title("LaMa Infill")
@tags("image", "inpaint")
@invocation("infill_lama", title="LaMa Infill", tags=["image", "inpaint"], category="inpaint", version="1.0.0")
class LaMaInfillInvocation(BaseInvocation):
"""Infills transparent areas of an image using the LaMa model"""
type: Literal["infill_lama"] = "infill_lama"
# Inputs
image: ImageField = InputField(description="The image to infill")
def invoke(self, context: InvocationContext) -> ImageOutput:

271
invokeai/app/invocations/latent.py
View File

@@ -21,6 +21,8 @@ from torchvision.transforms.functional import resize as tv_resize
from invokeai.app.invocations.metadata import CoreMetadata
from invokeai.app.invocations.primitives import (
DenoiseMaskField,
DenoiseMaskOutput,
ImageField,
ImageOutput,
LatentsField,
@@ -31,8 +33,9 @@ from invokeai.app.util.controlnet_utils import prepare_control_image
from invokeai.app.util.step_callback import stable_diffusion_step_callback
from invokeai.backend.model_management.models import ModelType, SilenceWarnings
from ...backend.model_management.models import BaseModelType
from ...backend.model_management.lora import ModelPatcher
from ...backend.model_management.seamless import set_seamless
from ...backend.model_management.models import BaseModelType
from ...backend.stable_diffusion import PipelineIntermediateState
from ...backend.stable_diffusion.diffusers_pipeline import (
ConditioningData,
@@ -46,13 +49,15 @@ from ...backend.util.devices import choose_precision, choose_torch_device
from ..models.image import ImageCategory, ResourceOrigin
from .baseinvocation import (
BaseInvocation,
BaseInvocationOutput,
FieldDescriptions,
Input,
InputField,
InvocationContext,
OutputField,
UIType,
tags,
title,
invocation,
invocation_output,
)
from .compel import ConditioningField
from .controlnet_image_processors import ControlField
@@ -64,6 +69,86 @@ DEFAULT_PRECISION = choose_precision(choose_torch_device())
SAMPLER_NAME_VALUES = Literal[tuple(list(SCHEDULER_MAP.keys()))]
@invocation_output("scheduler_output")
class SchedulerOutput(BaseInvocationOutput):
scheduler: SAMPLER_NAME_VALUES = OutputField(description=FieldDescriptions.scheduler, ui_type=UIType.Scheduler)
@invocation("scheduler", title="Scheduler", tags=["scheduler"], category="latents", version="1.0.0")
class SchedulerInvocation(BaseInvocation):
"""Selects a scheduler."""
scheduler: SAMPLER_NAME_VALUES = InputField(
default="euler", description=FieldDescriptions.scheduler, ui_type=UIType.Scheduler
)
def invoke(self, context: InvocationContext) -> SchedulerOutput:
return SchedulerOutput(scheduler=self.scheduler)
@invocation(
"create_denoise_mask", title="Create Denoise Mask", tags=["mask", "denoise"], category="latents", version="1.0.0"
)
class CreateDenoiseMaskInvocation(BaseInvocation):
"""Creates mask for denoising model run."""
vae: VaeField = InputField(description=FieldDescriptions.vae, input=Input.Connection, ui_order=0)
image: Optional[ImageField] = InputField(default=None, description="Image which will be masked", ui_order=1)
mask: ImageField = InputField(description="The mask to use when pasting", ui_order=2)
tiled: bool = InputField(default=False, description=FieldDescriptions.tiled, ui_order=3)
fp32: bool = InputField(default=DEFAULT_PRECISION == "float32", description=FieldDescriptions.fp32, ui_order=4)
def prep_mask_tensor(self, mask_image):
if mask_image.mode != "L":
mask_image = mask_image.convert("L")
mask_tensor = image_resized_to_grid_as_tensor(mask_image, normalize=False)
if mask_tensor.dim() == 3:
mask_tensor = mask_tensor.unsqueeze(0)
# if shape is not None:
# mask_tensor = tv_resize(mask_tensor, shape, T.InterpolationMode.BILINEAR)
return mask_tensor
@torch.no_grad()
def invoke(self, context: InvocationContext) -> DenoiseMaskOutput:
if self.image is not None:
image = context.services.images.get_pil_image(self.image.image_name)
image = image_resized_to_grid_as_tensor(image.convert("RGB"))
if image.dim() == 3:
image = image.unsqueeze(0)
else:
image = None
mask = self.prep_mask_tensor(
context.services.images.get_pil_image(self.mask.image_name),
)
if image is not None:
vae_info = context.services.model_manager.get_model(
**self.vae.vae.dict(),
context=context,
)
img_mask = tv_resize(mask, image.shape[-2:], T.InterpolationMode.BILINEAR, antialias=False)
masked_image = image * torch.where(img_mask < 0.5, 0.0, 1.0)
# TODO:
masked_latents = ImageToLatentsInvocation.vae_encode(vae_info, self.fp32, self.tiled, masked_image.clone())
masked_latents_name = f"{context.graph_execution_state_id}__{self.id}_masked_latents"
context.services.latents.save(masked_latents_name, masked_latents)
else:
masked_latents_name = None
mask_name = f"{context.graph_execution_state_id}__{self.id}_mask"
context.services.latents.save(mask_name, mask)
return DenoiseMaskOutput(
denoise_mask=DenoiseMaskField(
mask_name=mask_name,
masked_latents_name=masked_latents_name,
),
)
def get_scheduler(
context: InvocationContext,
scheduler_info: ModelInfo,
@@ -98,14 +183,16 @@ def get_scheduler(
return scheduler
@title("Denoise Latents")
@tags("latents", "denoise", "txt2img", "t2i", "t2l", "img2img", "i2i", "l2l")
@invocation(
"denoise_latents",
title="Denoise Latents",
tags=["latents", "denoise", "txt2img", "t2i", "t2l", "img2img", "i2i", "l2l"],
category="latents",
version="1.0.0",
)
class DenoiseLatentsInvocation(BaseInvocation):
"""Denoises noisy latents to decodable images"""
type: Literal["denoise_latents"] = "denoise_latents"
# Inputs
positive_conditioning: ConditioningField = InputField(
description=FieldDescriptions.positive_cond, input=Input.Connection, ui_order=0
)
@@ -124,14 +211,14 @@ class DenoiseLatentsInvocation(BaseInvocation):
)
unet: UNetField = InputField(description=FieldDescriptions.unet, input=Input.Connection, title="UNet", ui_order=2)
control: Union[ControlField, list[ControlField]] = InputField(
default=None, description=FieldDescriptions.control, input=Input.Connection, ui_order=5
)
latents: Optional[LatentsField] = InputField(
description=FieldDescriptions.latents, input=Input.Connection, ui_order=4
)
mask: Optional[ImageField] = InputField(
default=None,
description=FieldDescriptions.mask,
description=FieldDescriptions.control,
input=Input.Connection,
ui_order=5,
)
latents: Optional[LatentsField] = InputField(description=FieldDescriptions.latents, input=Input.Connection)
denoise_mask: Optional[DenoiseMaskField] = InputField(
default=None, description=FieldDescriptions.mask, input=Input.Connection, ui_order=6
)
@validator("cfg_scale")
@@ -235,7 +322,7 @@ class DenoiseLatentsInvocation(BaseInvocation):
context: InvocationContext,
# really only need model for dtype and device
model: StableDiffusionGeneratorPipeline,
control_input: List[ControlField],
control_input: Union[ControlField, List[ControlField]],
latents_shape: List[int],
exit_stack: ExitStack,
do_classifier_free_guidance: bool = True,
@@ -309,52 +396,46 @@ class DenoiseLatentsInvocation(BaseInvocation):
# original idea by https://github.com/AmericanPresidentJimmyCarter
# TODO: research more for second order schedulers timesteps
def init_scheduler(self, scheduler, device, steps, denoising_start, denoising_end):
num_inference_steps = steps
if scheduler.config.get("cpu_only", False):
scheduler.set_timesteps(num_inference_steps, device="cpu")
scheduler.set_timesteps(steps, device="cpu")
timesteps = scheduler.timesteps.to(device=device)
else:
scheduler.set_timesteps(num_inference_steps, device=device)
scheduler.set_timesteps(steps, device=device)
timesteps = scheduler.timesteps
# apply denoising_start
# skip greater order timesteps
_timesteps = timesteps[:: scheduler.order]
# get start timestep index
t_start_val = int(round(scheduler.config.num_train_timesteps * (1 - denoising_start)))
t_start_idx = len(list(filter(lambda ts: ts >= t_start_val, timesteps)))
timesteps = timesteps[t_start_idx:]
if scheduler.order == 2 and t_start_idx > 0:
timesteps = timesteps[1:]
t_start_idx = len(list(filter(lambda ts: ts >= t_start_val, _timesteps)))
# save start timestep to apply noise
init_timestep = timesteps[:1]
# apply denoising_end
# get end timestep index
t_end_val = int(round(scheduler.config.num_train_timesteps * (1 - denoising_end)))
t_end_idx = len(list(filter(lambda ts: ts >= t_end_val, timesteps)))
if scheduler.order == 2 and t_end_idx > 0:
t_end_idx += 1
timesteps = timesteps[:t_end_idx]
t_end_idx = len(list(filter(lambda ts: ts >= t_end_val, _timesteps[t_start_idx:])))
# calculate step count based on scheduler order
num_inference_steps = len(timesteps)
if scheduler.order == 2:
num_inference_steps += num_inference_steps % 2
num_inference_steps = num_inference_steps // 2
# apply order to indexes
t_start_idx *= scheduler.order
t_end_idx *= scheduler.order
init_timestep = timesteps[t_start_idx : t_start_idx + 1]
timesteps = timesteps[t_start_idx : t_start_idx + t_end_idx]
num_inference_steps = len(timesteps) // scheduler.order
return num_inference_steps, timesteps, init_timestep
def prep_mask_tensor(self, mask, context, lantents):
if mask is None:
return None
def prep_inpaint_mask(self, context, latents):
if self.denoise_mask is None:
return None, None
mask_image = context.services.images.get_pil_image(mask.image_name)
if mask_image.mode != "L":
# FIXME: why do we get passed an RGB image here? We can only use single-channel.
mask_image = mask_image.convert("L")
mask_tensor = image_resized_to_grid_as_tensor(mask_image, normalize=False)
if mask_tensor.dim() == 3:
mask_tensor = mask_tensor.unsqueeze(0)
mask_tensor = tv_resize(mask_tensor, lantents.shape[-2:], T.InterpolationMode.BILINEAR)
return 1 - mask_tensor
mask = context.services.latents.get(self.denoise_mask.mask_name)
mask = tv_resize(mask, latents.shape[-2:], T.InterpolationMode.BILINEAR, antialias=False)
if self.denoise_mask.masked_latents_name is not None:
masked_latents = context.services.latents.get(self.denoise_mask.masked_latents_name)
else:
masked_latents = None
return 1 - mask, masked_latents
@torch.no_grad()
def invoke(self, context: InvocationContext) -> LatentsOutput:
@@ -369,13 +450,19 @@ class DenoiseLatentsInvocation(BaseInvocation):
latents = context.services.latents.get(self.latents.latents_name)
if seed is None:
seed = self.latents.seed
else:
if noise is not None and noise.shape[1:] != latents.shape[1:]:
raise Exception(f"Incompatable 'noise' and 'latents' shapes: {latents.shape=} {noise.shape=}")
elif noise is not None:
latents = torch.zeros_like(noise)
else:
raise Exception("'latents' or 'noise' must be provided!")
if seed is None:
seed = 0
mask = self.prep_mask_tensor(self.mask, context, latents)
mask, masked_latents = self.prep_inpaint_mask(context, latents)
# Get the source node id (we are invoking the prepared node)
graph_execution_state = context.services.graph_execution_manager.get(context.graph_execution_state_id)
@@ -400,12 +487,14 @@ class DenoiseLatentsInvocation(BaseInvocation):
)
with ExitStack() as exit_stack, ModelPatcher.apply_lora_unet(
unet_info.context.model, _lora_loader()
), unet_info as unet:
), set_seamless(unet_info.context.model, self.unet.seamless_axes), unet_info as unet:
latents = latents.to(device=unet.device, dtype=unet.dtype)
if noise is not None:
noise = noise.to(device=unet.device, dtype=unet.dtype)
if mask is not None:
mask = mask.to(device=unet.device, dtype=unet.dtype)
if masked_latents is not None:
masked_latents = masked_latents.to(device=unet.device, dtype=unet.dtype)
scheduler = get_scheduler(
context=context,
@@ -442,6 +531,7 @@ class DenoiseLatentsInvocation(BaseInvocation):
noise=noise,
seed=seed,
mask=mask,
masked_latents=masked_latents,
num_inference_steps=num_inference_steps,
conditioning_data=conditioning_data,
control_data=control_data, # list[ControlNetData]
@@ -457,14 +547,12 @@ class DenoiseLatentsInvocation(BaseInvocation):
return build_latents_output(latents_name=name, latents=result_latents, seed=seed)
@title("Latents to Image")
@tags("latents", "image", "vae", "l2i")
@invocation(
"l2i", title="Latents to Image", tags=["latents", "image", "vae", "l2i"], category="latents", version="1.0.0"
)
class LatentsToImageInvocation(BaseInvocation):
"""Generates an image from latents."""
type: Literal["l2i"] = "l2i"
# Inputs
latents: LatentsField = InputField(
description=FieldDescriptions.latents,
input=Input.Connection,
@@ -490,7 +578,7 @@ class LatentsToImageInvocation(BaseInvocation):
context=context,
)
with vae_info as vae:
with set_seamless(vae_info.context.model, self.vae.seamless_axes), vae_info as vae:
latents = latents.to(vae.device)
if self.fp32:
vae.to(dtype=torch.float32)
@@ -545,6 +633,7 @@ class LatentsToImageInvocation(BaseInvocation):
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
metadata=self.metadata.dict() if self.metadata else None,
workflow=self.workflow,
)
return ImageOutput(
@@ -557,14 +646,10 @@ class LatentsToImageInvocation(BaseInvocation):
LATENTS_INTERPOLATION_MODE = Literal["nearest", "linear", "bilinear", "bicubic", "trilinear", "area", "nearest-exact"]
@title("Resize Latents")
@tags("latents", "resize")
@invocation("lresize", title="Resize Latents", tags=["latents", "resize"], category="latents", version="1.0.0")
class ResizeLatentsInvocation(BaseInvocation):
"""Resizes latents to explicit width/height (in pixels). Provided dimensions are floor-divided by 8."""
type: Literal["lresize"] = "lresize"
# Inputs
latents: LatentsField = InputField(
description=FieldDescriptions.latents,
input=Input.Connection,
@@ -605,14 +690,10 @@ class ResizeLatentsInvocation(BaseInvocation):
return build_latents_output(latents_name=name, latents=resized_latents, seed=self.latents.seed)
@title("Scale Latents")
@tags("latents", "resize")
@invocation("lscale", title="Scale Latents", tags=["latents", "resize"], category="latents", version="1.0.0")
class ScaleLatentsInvocation(BaseInvocation):
"""Scales latents by a given factor."""
type: Literal["lscale"] = "lscale"
# Inputs
latents: LatentsField = InputField(
description=FieldDescriptions.latents,
input=Input.Connection,
@@ -645,14 +726,12 @@ class ScaleLatentsInvocation(BaseInvocation):
return build_latents_output(latents_name=name, latents=resized_latents, seed=self.latents.seed)
@title("Image to Latents")
@tags("latents", "image", "vae", "i2l")
@invocation(
"i2l", title="Image to Latents", tags=["latents", "image", "vae", "i2l"], category="latents", version="1.0.0"
)
class ImageToLatentsInvocation(BaseInvocation):
"""Encodes an image into latents."""
type: Literal["i2l"] = "i2l"
# Inputs
image: ImageField = InputField(
description="The image to encode",
)
@@ -663,26 +742,11 @@ class ImageToLatentsInvocation(BaseInvocation):
tiled: bool = InputField(default=False, description=FieldDescriptions.tiled)
fp32: bool = InputField(default=DEFAULT_PRECISION == "float32", description=FieldDescriptions.fp32)
@torch.no_grad()
def invoke(self, context: InvocationContext) -> LatentsOutput:
# image = context.services.images.get(
# self.image.image_type, self.image.image_name
# )
image = context.services.images.get_pil_image(self.image.image_name)
# vae_info = context.services.model_manager.get_model(**self.vae.vae.dict())
vae_info = context.services.model_manager.get_model(
**self.vae.vae.dict(),
context=context,
)
image_tensor = image_resized_to_grid_as_tensor(image.convert("RGB"))
if image_tensor.dim() == 3:
image_tensor = einops.rearrange(image_tensor, "c h w -> 1 c h w")
@staticmethod
def vae_encode(vae_info, upcast, tiled, image_tensor):
with vae_info as vae:
orig_dtype = vae.dtype
if self.fp32:
if upcast:
vae.to(dtype=torch.float32)
use_torch_2_0_or_xformers = isinstance(
@@ -707,7 +771,7 @@ class ImageToLatentsInvocation(BaseInvocation):
vae.to(dtype=torch.float16)
# latents = latents.half()
if self.tiled:
if tiled:
vae.enable_tiling()
else:
vae.disable_tiling()
@@ -721,20 +785,33 @@ class ImageToLatentsInvocation(BaseInvocation):
latents = vae.config.scaling_factor * latents
latents = latents.to(dtype=orig_dtype)
return latents
@torch.no_grad()
def invoke(self, context: InvocationContext) -> LatentsOutput:
image = context.services.images.get_pil_image(self.image.image_name)
vae_info = context.services.model_manager.get_model(
**self.vae.vae.dict(),
context=context,
)
image_tensor = image_resized_to_grid_as_tensor(image.convert("RGB"))
if image_tensor.dim() == 3:
image_tensor = einops.rearrange(image_tensor, "c h w -> 1 c h w")
latents = self.vae_encode(vae_info, self.fp32, self.tiled, image_tensor)
name = f"{context.graph_execution_state_id}__{self.id}"
latents = latents.to("cpu")
context.services.latents.save(name, latents)
return build_latents_output(latents_name=name, latents=latents, seed=None)
@title("Blend Latents")
@tags("latents", "blend")
@invocation("lblend", title="Blend Latents", tags=["latents", "blend"], category="latents", version="1.0.0")
class BlendLatentsInvocation(BaseInvocation):
"""Blend two latents using a given alpha. Latents must have same size."""
type: Literal["lblend"] = "lblend"
# Inputs
latents_a: LatentsField = InputField(
description=FieldDescriptions.latents,
input=Input.Connection,

34
invokeai/app/invocations/math.py
View File

@@ -1,22 +1,16 @@
# Copyright (c) 2023 Kyle Schouviller (https://github.com/kyle0654)
from typing import Literal
import numpy as np
from invokeai.app.invocations.primitives import IntegerOutput
from .baseinvocation import BaseInvocation, FieldDescriptions, InputField, InvocationContext, tags, title
from .baseinvocation import BaseInvocation, FieldDescriptions, InputField, InvocationContext, invocation
@title("Add Integers")
@tags("math")
@invocation("add", title="Add Integers", tags=["math", "add"], category="math", version="1.0.0")
class AddInvocation(BaseInvocation):
"""Adds two numbers"""
type: Literal["add"] = "add"
# Inputs
a: int = InputField(default=0, description=FieldDescriptions.num_1)
b: int = InputField(default=0, description=FieldDescriptions.num_2)
@@ -24,14 +18,10 @@ class AddInvocation(BaseInvocation):
return IntegerOutput(value=self.a + self.b)
@title("Subtract Integers")
@tags("math")
@invocation("sub", title="Subtract Integers", tags=["math", "subtract"], category="math", version="1.0.0")
class SubtractInvocation(BaseInvocation):
"""Subtracts two numbers"""
type: Literal["sub"] = "sub"
# Inputs
a: int = InputField(default=0, description=FieldDescriptions.num_1)
b: int = InputField(default=0, description=FieldDescriptions.num_2)
@@ -39,14 +29,10 @@ class SubtractInvocation(BaseInvocation):
return IntegerOutput(value=self.a - self.b)
@title("Multiply Integers")
@tags("math")
@invocation("mul", title="Multiply Integers", tags=["math", "multiply"], category="math", version="1.0.0")
class MultiplyInvocation(BaseInvocation):
"""Multiplies two numbers"""
type: Literal["mul"] = "mul"
# Inputs
a: int = InputField(default=0, description=FieldDescriptions.num_1)
b: int = InputField(default=0, description=FieldDescriptions.num_2)
@@ -54,14 +40,10 @@ class MultiplyInvocation(BaseInvocation):
return IntegerOutput(value=self.a * self.b)
@title("Divide Integers")
@tags("math")
@invocation("div", title="Divide Integers", tags=["math", "divide"], category="math", version="1.0.0")
class DivideInvocation(BaseInvocation):
"""Divides two numbers"""
type: Literal["div"] = "div"
# Inputs
a: int = InputField(default=0, description=FieldDescriptions.num_1)
b: int = InputField(default=0, description=FieldDescriptions.num_2)
@@ -69,14 +51,10 @@ class DivideInvocation(BaseInvocation):
return IntegerOutput(value=int(self.a / self.b))
@title("Random Integer")
@tags("math")
@invocation("rand_int", title="Random Integer", tags=["math", "random"], category="math", version="1.0.0")
class RandomIntInvocation(BaseInvocation):
"""Outputs a single random integer."""
type: Literal["rand_int"] = "rand_int"
# Inputs
low: int = InputField(default=0, description="The inclusive low value")
high: int = InputField(default=np.iinfo(np.int32).max, description="The exclusive high value")

24
invokeai/app/invocations/metadata.py
View File

@@ -1,4 +1,4 @@
from typing import Literal, Optional
from typing import Optional
from pydantic import Field
@@ -8,8 +8,8 @@ from invokeai.app.invocations.baseinvocation import (
InputField,
InvocationContext,
OutputField,
tags,
title,
invocation,
invocation_output,
)
from invokeai.app.invocations.controlnet_image_processors import ControlField
from invokeai.app.invocations.model import LoRAModelField, MainModelField, VAEModelField
@@ -72,10 +72,10 @@ class CoreMetadata(BaseModelExcludeNull):
)
refiner_steps: Optional[int] = Field(default=None, description="The number of steps used for the refiner")
refiner_scheduler: Optional[str] = Field(default=None, description="The scheduler used for the refiner")
refiner_positive_aesthetic_store: Optional[float] = Field(
refiner_positive_aesthetic_score: Optional[float] = Field(
default=None, description="The aesthetic score used for the refiner"
)
refiner_negative_aesthetic_store: Optional[float] = Field(
refiner_negative_aesthetic_score: Optional[float] = Field(
default=None, description="The aesthetic score used for the refiner"
)
refiner_start: Optional[float] = Field(default=None, description="The start value used for refiner denoising")
@@ -91,21 +91,19 @@ class ImageMetadata(BaseModelExcludeNull):
graph: Optional[dict] = Field(default=None, description="The graph that created the image")
@invocation_output("metadata_accumulator_output")
class MetadataAccumulatorOutput(BaseInvocationOutput):
"""The output of the MetadataAccumulator node"""
type: Literal["metadata_accumulator_output"] = "metadata_accumulator_output"
metadata: CoreMetadata = OutputField(description="The core metadata for the image")
@title("Metadata Accumulator")
@tags("metadata")
@invocation(
"metadata_accumulator", title="Metadata Accumulator", tags=["metadata"], category="metadata", version="1.0.0"
)
class MetadataAccumulatorInvocation(BaseInvocation):
"""Outputs a Core Metadata Object"""
type: Literal["metadata_accumulator"] = "metadata_accumulator"
generation_mode: str = InputField(
description="The generation mode that output this image",
)
@@ -164,11 +162,11 @@ class MetadataAccumulatorInvocation(BaseInvocation):
default=None,
description="The scheduler used for the refiner",
)
refiner_positive_aesthetic_store: Optional[float] = InputField(
refiner_positive_aesthetic_score: Optional[float] = InputField(
default=None,
description="The aesthetic score used for the refiner",
)
refiner_negative_aesthetic_store: Optional[float] = InputField(
refiner_negative_aesthetic_score: Optional[float] = InputField(
default=None,
description="The aesthetic score used for the refiner",
)

103
invokeai/app/invocations/model.py
View File

@@ -1,5 +1,5 @@
import copy
from typing import List, Literal, Optional
from typing import List, Optional
from pydantic import BaseModel, Field
@@ -8,13 +8,13 @@ from .baseinvocation import (
BaseInvocation,
BaseInvocationOutput,
FieldDescriptions,
InputField,
Input,
InputField,
InvocationContext,
OutputField,
UIType,
tags,
title,
invocation,
invocation_output,
)
@@ -33,6 +33,7 @@ class UNetField(BaseModel):
unet: ModelInfo = Field(description="Info to load unet submodel")
scheduler: ModelInfo = Field(description="Info to load scheduler submodel")
loras: List[LoraInfo] = Field(description="Loras to apply on model loading")
seamless_axes: List[str] = Field(default_factory=list, description='Axes("x" and "y") to which apply seamless')
class ClipField(BaseModel):
@@ -45,13 +46,13 @@ class ClipField(BaseModel):
class VaeField(BaseModel):
# TODO: better naming?
vae: ModelInfo = Field(description="Info to load vae submodel")
seamless_axes: List[str] = Field(default_factory=list, description='Axes("x" and "y") to which apply seamless')
@invocation_output("model_loader_output")
class ModelLoaderOutput(BaseInvocationOutput):
"""Model loader output"""
type: Literal["model_loader_output"] = "model_loader_output"
unet: UNetField = OutputField(description=FieldDescriptions.unet, title="UNet")
clip: ClipField = OutputField(description=FieldDescriptions.clip, title="CLIP")
vae: VaeField = OutputField(description=FieldDescriptions.vae, title="VAE")
@@ -72,14 +73,10 @@ class LoRAModelField(BaseModel):
base_model: BaseModelType = Field(description="Base model")
@title("Main Model")
@tags("model")
@invocation("main_model_loader", title="Main Model", tags=["model"], category="model", version="1.0.0")
class MainModelLoaderInvocation(BaseInvocation):
"""Loads a main model, outputting its submodels."""
type: Literal["main_model_loader"] = "main_model_loader"
# Inputs
model: MainModelField = InputField(description=FieldDescriptions.main_model, input=Input.Direct)
# TODO: precision?
@@ -168,25 +165,18 @@ class MainModelLoaderInvocation(BaseInvocation):
)
@invocation_output("lora_loader_output")
class LoraLoaderOutput(BaseInvocationOutput):
"""Model loader output"""
# fmt: off
type: Literal["lora_loader_output"] = "lora_loader_output"
unet: Optional[UNetField] = OutputField(default=None, description=FieldDescriptions.unet, title="UNet")
clip: Optional[ClipField] = OutputField(default=None, description=FieldDescriptions.clip, title="CLIP")
# fmt: on
@title("LoRA")
@tags("lora", "model")
@invocation("lora_loader", title="LoRA", tags=["model"], category="model", version="1.0.0")
class LoraLoaderInvocation(BaseInvocation):
"""Apply selected lora to unet and text_encoder."""
type: Literal["lora_loader"] = "lora_loader"
# Inputs
lora: LoRAModelField = InputField(description=FieldDescriptions.lora_model, input=Input.Direct, title="LoRA")
weight: float = InputField(default=0.75, description=FieldDescriptions.lora_weight)
unet: Optional[UNetField] = InputField(
@@ -245,34 +235,28 @@ class LoraLoaderInvocation(BaseInvocation):
return output
@invocation_output("sdxl_lora_loader_output")
class SDXLLoraLoaderOutput(BaseInvocationOutput):
"""SDXL LoRA Loader Output"""
# fmt: off
type: Literal["sdxl_lora_loader_output"] = "sdxl_lora_loader_output"
unet: Optional[UNetField] = OutputField(default=None, description=FieldDescriptions.unet, title="UNet")
clip: Optional[ClipField] = OutputField(default=None, description=FieldDescriptions.clip, title="CLIP 1")
clip2: Optional[ClipField] = OutputField(default=None, description=FieldDescriptions.clip, title="CLIP 2")
# fmt: on
@title("SDXL LoRA")
@tags("sdxl", "lora", "model")
@invocation("sdxl_lora_loader", title="SDXL LoRA", tags=["lora", "model"], category="model", version="1.0.0")
class SDXLLoraLoaderInvocation(BaseInvocation):
"""Apply selected lora to unet and text_encoder."""
type: Literal["sdxl_lora_loader"] = "sdxl_lora_loader"
lora: LoRAModelField = InputField(description=FieldDescriptions.lora_model, input=Input.Direct, title="LoRA")
weight: float = Field(default=0.75, description=FieldDescriptions.lora_weight)
unet: Optional[UNetField] = Field(
default=None, description=FieldDescriptions.unet, input=Input.Connection, title="UNET"
weight: float = InputField(default=0.75, description=FieldDescriptions.lora_weight)
unet: Optional[UNetField] = InputField(
default=None, description=FieldDescriptions.unet, input=Input.Connection, title="UNet"
)
clip: Optional[ClipField] = Field(
clip: Optional[ClipField] = InputField(
default=None, description=FieldDescriptions.clip, input=Input.Connection, title="CLIP 1"
)
clip2: Optional[ClipField] = Field(
clip2: Optional[ClipField] = InputField(
default=None, description=FieldDescriptions.clip, input=Input.Connection, title="CLIP 2"
)
@@ -347,23 +331,17 @@ class VAEModelField(BaseModel):
base_model: BaseModelType = Field(description="Base model")
@invocation_output("vae_loader_output")
class VaeLoaderOutput(BaseInvocationOutput):
"""Model loader output"""
"""VAE output"""
type: Literal["vae_loader_output"] = "vae_loader_output"
# Outputs
vae: VaeField = OutputField(description=FieldDescriptions.vae, title="VAE")
@title("VAE")
@tags("vae", "model")
@invocation("vae_loader", title="VAE", tags=["vae", "model"], category="model", version="1.0.0")
class VaeLoaderInvocation(BaseInvocation):
"""Loads a VAE model, outputting a VaeLoaderOutput"""
type: Literal["vae_loader"] = "vae_loader"
# Inputs
vae_model: VAEModelField = InputField(
description=FieldDescriptions.vae_model, input=Input.Direct, ui_type=UIType.VaeModel, title="VAE"
)
@@ -388,3 +366,44 @@ class VaeLoaderInvocation(BaseInvocation):
)
)
)
@invocation_output("seamless_output")
class SeamlessModeOutput(BaseInvocationOutput):
"""Modified Seamless Model output"""
unet: Optional[UNetField] = OutputField(description=FieldDescriptions.unet, title="UNet")
vae: Optional[VaeField] = OutputField(description=FieldDescriptions.vae, title="VAE")
@invocation("seamless", title="Seamless", tags=["seamless", "model"], category="model", version="1.0.0")
class SeamlessModeInvocation(BaseInvocation):
"""Applies the seamless transformation to the Model UNet and VAE."""
unet: Optional[UNetField] = InputField(
default=None, description=FieldDescriptions.unet, input=Input.Connection, title="UNet"
)
vae: Optional[VaeField] = InputField(
default=None, description=FieldDescriptions.vae_model, input=Input.Connection, title="VAE"
)
seamless_y: bool = InputField(default=True, input=Input.Any, description="Specify whether Y axis is seamless")
seamless_x: bool = InputField(default=True, input=Input.Any, description="Specify whether X axis is seamless")
def invoke(self, context: InvocationContext) -> SeamlessModeOutput:
# Conditionally append 'x' and 'y' based on seamless_x and seamless_y
unet = copy.deepcopy(self.unet)
vae = copy.deepcopy(self.vae)
seamless_axes_list = []
if self.seamless_x:
seamless_axes_list.append("x")
if self.seamless_y:
seamless_axes_list.append("y")
if unet is not None:
unet.seamless_axes = seamless_axes_list
if vae is not None:
vae.seamless_axes = seamless_axes_list
return SeamlessModeOutput(unet=unet, vae=vae)

15
invokeai/app/invocations/noise.py
View File

@@ -1,6 +1,5 @@
# Copyright (c) 2023 Kyle Schouviller (https://github.com/kyle0654) & the InvokeAI Team
from typing import Literal
import torch
from pydantic import validator
@@ -16,8 +15,8 @@ from .baseinvocation import (
InputField,
InvocationContext,
OutputField,
tags,
title,
invocation,
invocation_output,
)
"""
@@ -62,12 +61,10 @@ Nodes
"""
@invocation_output("noise_output")
class NoiseOutput(BaseInvocationOutput):
"""Invocation noise output"""
type: Literal["noise_output"] = "noise_output"
# Inputs
noise: LatentsField = OutputField(default=None, description=FieldDescriptions.noise)
width: int = OutputField(description=FieldDescriptions.width)
height: int = OutputField(description=FieldDescriptions.height)
@@ -81,14 +78,10 @@ def build_noise_output(latents_name: str, latents: torch.Tensor, seed: int):
)
@title("Noise")
@tags("latents", "noise")
@invocation("noise", title="Noise", tags=["latents", "noise"], category="latents", version="1.0.0")
class NoiseInvocation(BaseInvocation):
"""Generates latent noise."""
type: Literal["noise"] = "noise"
# Inputs
seed: int = InputField(
ge=0,
le=SEED_MAX,

45
invokeai/app/invocations/onnx.py
View File

@@ -31,8 +31,8 @@ from .baseinvocation import (
OutputField,
UIComponent,
UIType,
tags,
title,
invocation,
invocation_output,
)
from .controlnet_image_processors import ControlField
from .latent import SAMPLER_NAME_VALUES, LatentsField, LatentsOutput, build_latents_output, get_scheduler
@@ -56,11 +56,8 @@ ORT_TO_NP_TYPE = {
PRECISION_VALUES = Literal[tuple(list(ORT_TO_NP_TYPE.keys()))]
@title("ONNX Prompt (Raw)")
@tags("onnx", "prompt")
@invocation("prompt_onnx", title="ONNX Prompt (Raw)", tags=["prompt", "onnx"], category="conditioning", version="1.0.0")
class ONNXPromptInvocation(BaseInvocation):
type: Literal["prompt_onnx"] = "prompt_onnx"
prompt: str = InputField(default="", description=FieldDescriptions.raw_prompt, ui_component=UIComponent.Textarea)
clip: ClipField = InputField(description=FieldDescriptions.clip, input=Input.Connection)
@@ -141,14 +138,16 @@ class ONNXPromptInvocation(BaseInvocation):
# Text to image
@title("ONNX Text to Latents")
@tags("latents", "inference", "txt2img", "onnx")
@invocation(
"t2l_onnx",
title="ONNX Text to Latents",
tags=["latents", "inference", "txt2img", "onnx"],
category="latents",
version="1.0.0",
)
class ONNXTextToLatentsInvocation(BaseInvocation):
"""Generates latents from conditionings."""
type: Literal["t2l_onnx"] = "t2l_onnx"
# Inputs
positive_conditioning: ConditioningField = InputField(
description=FieldDescriptions.positive_cond,
input=Input.Connection,
@@ -316,14 +315,16 @@ class ONNXTextToLatentsInvocation(BaseInvocation):
# Latent to image
@title("ONNX Latents to Image")
@tags("latents", "image", "vae", "onnx")
@invocation(
"l2i_onnx",
title="ONNX Latents to Image",
tags=["latents", "image", "vae", "onnx"],
category="image",
version="1.0.0",
)
class ONNXLatentsToImageInvocation(BaseInvocation):
"""Generates an image from latents."""
type: Literal["l2i_onnx"] = "l2i_onnx"
# Inputs
latents: LatentsField = InputField(
description=FieldDescriptions.denoised_latents,
input=Input.Connection,
@@ -376,6 +377,7 @@ class ONNXLatentsToImageInvocation(BaseInvocation):
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
metadata=self.metadata.dict() if self.metadata else None,
workflow=self.workflow,
)
return ImageOutput(
@@ -385,17 +387,14 @@ class ONNXLatentsToImageInvocation(BaseInvocation):
)
@invocation_output("model_loader_output_onnx")
class ONNXModelLoaderOutput(BaseInvocationOutput):
"""Model loader output"""
# fmt: off
type: Literal["model_loader_output_onnx"] = "model_loader_output_onnx"
unet: UNetField = OutputField(default=None, description=FieldDescriptions.unet, title="UNet")
clip: ClipField = OutputField(default=None, description=FieldDescriptions.clip, title="CLIP")
vae_decoder: VaeField = OutputField(default=None, description=FieldDescriptions.vae, title="VAE Decoder")
vae_encoder: VaeField = OutputField(default=None, description=FieldDescriptions.vae, title="VAE Encoder")
# fmt: on
class OnnxModelField(BaseModel):
@@ -406,14 +405,10 @@ class OnnxModelField(BaseModel):
model_type: ModelType = Field(description="Model Type")
@title("ONNX Main Model")
@tags("onnx", "model")
@invocation("onnx_model_loader", title="ONNX Main Model", tags=["onnx", "model"], category="model", version="1.0.0")
class OnnxModelLoaderInvocation(BaseInvocation):
"""Loads a main model, outputting its submodels."""
type: Literal["onnx_model_loader"] = "onnx_model_loader"
# Inputs
model: OnnxModelField = InputField(
description=FieldDescriptions.onnx_main_model, input=Input.Direct, ui_type=UIType.ONNXModel
)

14
invokeai/app/invocations/param_easing.py
View File

@@ -42,17 +42,13 @@ from matplotlib.ticker import MaxNLocator
from invokeai.app.invocations.primitives import FloatCollectionOutput
from .baseinvocation import BaseInvocation, InputField, InvocationContext, tags, title
from .baseinvocation import BaseInvocation, InputField, InvocationContext, invocation
@title("Float Range")
@tags("math", "range")
@invocation("float_range", title="Float Range", tags=["math", "range"], category="math", version="1.0.0")
class FloatLinearRangeInvocation(BaseInvocation):
"""Creates a range"""
type: Literal["float_range"] = "float_range"
# Inputs
start: float = InputField(default=5, description="The first value of the range")
stop: float = InputField(default=10, description="The last value of the range")
steps: int = InputField(default=30, description="number of values to interpolate over (including start and stop)")
@@ -100,14 +96,10 @@ EASING_FUNCTION_KEYS = Literal[tuple(list(EASING_FUNCTIONS_MAP.keys()))]
# actually I think for now could just use CollectionOutput (which is list[Any]
@title("Step Param Easing")
@tags("step", "easing")
@invocation("step_param_easing", title="Step Param Easing", tags=["step", "easing"], category="step", version="1.0.0")
class StepParamEasingInvocation(BaseInvocation):
"""Experimental per-step parameter easing for denoising steps"""
type: Literal["step_param_easing"] = "step_param_easing"
# Inputs
easing: EASING_FUNCTION_KEYS = InputField(default="Linear", description="The easing function to use")
num_steps: int = InputField(default=20, description="number of denoising steps")
start_value: float = InputField(default=0.0, description="easing starting value")

270
invokeai/app/invocations/primitives.py
View File

@@ -1,6 +1,6 @@
# Copyright (c) 2023 Kyle Schouviller (https://github.com/kyle0654)
from typing import Literal, Optional, Tuple
from typing import Optional, Tuple
import torch
from pydantic import BaseModel, Field
@@ -14,9 +14,8 @@ from .baseinvocation import (
InvocationContext,
OutputField,
UIComponent,
UIType,
tags,
title,
invocation,
invocation_output,
)
"""
@@ -29,47 +28,45 @@ Primitives: Boolean, Integer, Float, String, Image, Latents, Conditioning, Color
# region Boolean
@invocation_output("boolean_output")
class BooleanOutput(BaseInvocationOutput):
"""Base class for nodes that output a single boolean"""
type: Literal["boolean_output"] = "boolean_output"
value: bool = OutputField(description="The output boolean")
@invocation_output("boolean_collection_output")
class BooleanCollectionOutput(BaseInvocationOutput):
"""Base class for nodes that output a collection of booleans"""
type: Literal["boolean_collection_output"] = "boolean_collection_output"
# Outputs
collection: list[bool] = OutputField(description="The output boolean collection", ui_type=UIType.BooleanCollection)
collection: list[bool] = OutputField(
description="The output boolean collection",
)
@title("Boolean Primitive")
@tags("primitives", "boolean")
@invocation(
"boolean", title="Boolean Primitive", tags=["primitives", "boolean"], category="primitives", version="1.0.0"
)
class BooleanInvocation(BaseInvocation):
"""A boolean primitive value"""
type: Literal["boolean"] = "boolean"
# Inputs
value: bool = InputField(default=False, description="The boolean value")
def invoke(self, context: InvocationContext) -> BooleanOutput:
return BooleanOutput(value=self.value)
@title("Boolean Primitive Collection")
@tags("primitives", "boolean", "collection")
@invocation(
"boolean_collection",
title="Boolean Collection Primitive",
tags=["primitives", "boolean", "collection"],
category="primitives",
version="1.0.0",
)
class BooleanCollectionInvocation(BaseInvocation):
"""A collection of boolean primitive values"""
type: Literal["boolean_collection"] = "boolean_collection"
# Inputs
collection: list[bool] = InputField(
default_factory=list, description="The collection of boolean values", ui_type=UIType.BooleanCollection
)
collection: list[bool] = InputField(default_factory=list, description="The collection of boolean values")
def invoke(self, context: InvocationContext) -> BooleanCollectionOutput:
return BooleanCollectionOutput(collection=self.collection)
@@ -80,47 +77,45 @@ class BooleanCollectionInvocation(BaseInvocation):
# region Integer
@invocation_output("integer_output")
class IntegerOutput(BaseInvocationOutput):
"""Base class for nodes that output a single integer"""
type: Literal["integer_output"] = "integer_output"
value: int = OutputField(description="The output integer")
@invocation_output("integer_collection_output")
class IntegerCollectionOutput(BaseInvocationOutput):
"""Base class for nodes that output a collection of integers"""
type: Literal["integer_collection_output"] = "integer_collection_output"
# Outputs
collection: list[int] = OutputField(description="The int collection", ui_type=UIType.IntegerCollection)
collection: list[int] = OutputField(
description="The int collection",
)
@title("Integer Primitive")
@tags("primitives", "integer")
@invocation(
"integer", title="Integer Primitive", tags=["primitives", "integer"], category="primitives", version="1.0.0"
)
class IntegerInvocation(BaseInvocation):
"""An integer primitive value"""
type: Literal["integer"] = "integer"
# Inputs
value: int = InputField(default=0, description="The integer value")
def invoke(self, context: InvocationContext) -> IntegerOutput:
return IntegerOutput(value=self.value)
@title("Integer Primitive Collection")
@tags("primitives", "integer", "collection")
@invocation(
"integer_collection",
title="Integer Collection Primitive",
tags=["primitives", "integer", "collection"],
category="primitives",
version="1.0.0",
)
class IntegerCollectionInvocation(BaseInvocation):
"""A collection of integer primitive values"""
type: Literal["integer_collection"] = "integer_collection"
# Inputs
collection: list[int] = InputField(
default=0, description="The collection of integer values", ui_type=UIType.IntegerCollection
)
collection: list[int] = InputField(default_factory=list, description="The collection of integer values")
def invoke(self, context: InvocationContext) -> IntegerCollectionOutput:
return IntegerCollectionOutput(collection=self.collection)
@@ -131,47 +126,43 @@ class IntegerCollectionInvocation(BaseInvocation):
# region Float
@invocation_output("float_output")
class FloatOutput(BaseInvocationOutput):
"""Base class for nodes that output a single float"""
type: Literal["float_output"] = "float_output"
value: float = OutputField(description="The output float")
@invocation_output("float_collection_output")
class FloatCollectionOutput(BaseInvocationOutput):
"""Base class for nodes that output a collection of floats"""
type: Literal["float_collection_output"] = "float_collection_output"
# Outputs
collection: list[float] = OutputField(description="The float collection", ui_type=UIType.FloatCollection)
collection: list[float] = OutputField(
description="The float collection",
)
@title("Float Primitive")
@tags("primitives", "float")
@invocation("float", title="Float Primitive", tags=["primitives", "float"], category="primitives", version="1.0.0")
class FloatInvocation(BaseInvocation):
"""A float primitive value"""
type: Literal["float"] = "float"
# Inputs
value: float = InputField(default=0.0, description="The float value")
def invoke(self, context: InvocationContext) -> FloatOutput:
return FloatOutput(value=self.value)
@title("Float Primitive Collection")
@tags("primitives", "float", "collection")
@invocation(
"float_collection",
title="Float Collection Primitive",
tags=["primitives", "float", "collection"],
category="primitives",
version="1.0.0",
)
class FloatCollectionInvocation(BaseInvocation):
"""A collection of float primitive values"""
type: Literal["float_collection"] = "float_collection"
# Inputs
collection: list[float] = InputField(
default_factory=list, description="The collection of float values", ui_type=UIType.FloatCollection
)
collection: list[float] = InputField(default_factory=list, description="The collection of float values")
def invoke(self, context: InvocationContext) -> FloatCollectionOutput:
return FloatCollectionOutput(collection=self.collection)
@@ -182,47 +173,43 @@ class FloatCollectionInvocation(BaseInvocation):
# region String
@invocation_output("string_output")
class StringOutput(BaseInvocationOutput):
"""Base class for nodes that output a single string"""
type: Literal["string_output"] = "string_output"
value: str = OutputField(description="The output string")
@invocation_output("string_collection_output")
class StringCollectionOutput(BaseInvocationOutput):
"""Base class for nodes that output a collection of strings"""
type: Literal["string_collection_output"] = "string_collection_output"
# Outputs
collection: list[str] = OutputField(description="The output strings", ui_type=UIType.StringCollection)
collection: list[str] = OutputField(
description="The output strings",
)
@title("String Primitive")
@tags("primitives", "string")
@invocation("string", title="String Primitive", tags=["primitives", "string"], category="primitives", version="1.0.0")
class StringInvocation(BaseInvocation):
"""A string primitive value"""
type: Literal["string"] = "string"
# Inputs
value: str = InputField(default="", description="The string value", ui_component=UIComponent.Textarea)
def invoke(self, context: InvocationContext) -> StringOutput:
return StringOutput(value=self.value)
@title("String Primitive Collection")
@tags("primitives", "string", "collection")
@invocation(
"string_collection",
title="String Collection Primitive",
tags=["primitives", "string", "collection"],
category="primitives",
version="1.0.0",
)
class StringCollectionInvocation(BaseInvocation):
"""A collection of string primitive values"""
type: Literal["string_collection"] = "string_collection"
# Inputs
collection: list[str] = InputField(
default_factory=list, description="The collection of string values", ui_type=UIType.StringCollection
)
collection: list[str] = InputField(default_factory=list, description="The collection of string values")
def invoke(self, context: InvocationContext) -> StringCollectionOutput:
return StringCollectionOutput(collection=self.collection)
@@ -239,33 +226,28 @@ class ImageField(BaseModel):
image_name: str = Field(description="The name of the image")
@invocation_output("image_output")
class ImageOutput(BaseInvocationOutput):
"""Base class for nodes that output a single image"""
type: Literal["image_output"] = "image_output"
image: ImageField = OutputField(description="The output image")
width: int = OutputField(description="The width of the image in pixels")
height: int = OutputField(description="The height of the image in pixels")
@invocation_output("image_collection_output")
class ImageCollectionOutput(BaseInvocationOutput):
"""Base class for nodes that output a collection of images"""
type: Literal["image_collection_output"] = "image_collection_output"
# Outputs
collection: list[ImageField] = OutputField(description="The output images", ui_type=UIType.ImageCollection)
collection: list[ImageField] = OutputField(
description="The output images",
)
@title("Image Primitive")
@tags("primitives", "image")
@invocation("image", title="Image Primitive", tags=["primitives", "image"], category="primitives", version="1.0.0")
class ImageInvocation(BaseInvocation):
"""An image primitive value"""
# Metadata
type: Literal["image"] = "image"
# Inputs
image: ImageField = InputField(description="The image to load")
def invoke(self, context: InvocationContext) -> ImageOutput:
@@ -278,22 +260,41 @@ class ImageInvocation(BaseInvocation):
)
@title("Image Primitive Collection")
@tags("primitives", "image", "collection")
@invocation(
"image_collection",
title="Image Collection Primitive",
tags=["primitives", "image", "collection"],
category="primitives",
version="1.0.0",
)
class ImageCollectionInvocation(BaseInvocation):
"""A collection of image primitive values"""
type: Literal["image_collection"] = "image_collection"
# Inputs
collection: list[ImageField] = InputField(
default=0, description="The collection of image values", ui_type=UIType.ImageCollection
)
collection: list[ImageField] = InputField(description="The collection of image values")
def invoke(self, context: InvocationContext) -> ImageCollectionOutput:
return ImageCollectionOutput(collection=self.collection)
# endregion
# region DenoiseMask
class DenoiseMaskField(BaseModel):
"""An inpaint mask field"""
mask_name: str = Field(description="The name of the mask image")
masked_latents_name: Optional[str] = Field(description="The name of the masked image latents")
@invocation_output("denoise_mask_output")
class DenoiseMaskOutput(BaseInvocationOutput):
"""Base class for nodes that output a single image"""
denoise_mask: DenoiseMaskField = OutputField(description="Mask for denoise model run")
# endregion
# region Latents
@@ -306,11 +307,10 @@ class LatentsField(BaseModel):
seed: Optional[int] = Field(default=None, description="Seed used to generate this latents")
@invocation_output("latents_output")
class LatentsOutput(BaseInvocationOutput):
"""Base class for nodes that output a single latents tensor"""
type: Literal["latents_output"] = "latents_output"
latents: LatentsField = OutputField(
description=FieldDescriptions.latents,
)
@@ -318,25 +318,21 @@ class LatentsOutput(BaseInvocationOutput):
height: int = OutputField(description=FieldDescriptions.height)
@invocation_output("latents_collection_output")
class LatentsCollectionOutput(BaseInvocationOutput):
"""Base class for nodes that output a collection of latents tensors"""
type: Literal["latents_collection_output"] = "latents_collection_output"
collection: list[LatentsField] = OutputField(
description=FieldDescriptions.latents,
ui_type=UIType.LatentsCollection,
)
@title("Latents Primitive")
@tags("primitives", "latents")
@invocation(
"latents", title="Latents Primitive", tags=["primitives", "latents"], category="primitives", version="1.0.0"
)
class LatentsInvocation(BaseInvocation):
"""A latents tensor primitive value"""
type: Literal["latents"] = "latents"
# Inputs
latents: LatentsField = InputField(description="The latents tensor", input=Input.Connection)
def invoke(self, context: InvocationContext) -> LatentsOutput:
@@ -345,16 +341,18 @@ class LatentsInvocation(BaseInvocation):
return build_latents_output(self.latents.latents_name, latents)
@title("Latents Primitive Collection")
@tags("primitives", "latents", "collection")
@invocation(
"latents_collection",
title="Latents Collection Primitive",
tags=["primitives", "latents", "collection"],
category="primitives",
version="1.0.0",
)
class LatentsCollectionInvocation(BaseInvocation):
"""A collection of latents tensor primitive values"""
type: Literal["latents_collection"] = "latents_collection"
# Inputs
collection: list[LatentsField] = InputField(
description="The collection of latents tensors", ui_type=UIType.LatentsCollection
description="The collection of latents tensors",
)
def invoke(self, context: InvocationContext) -> LatentsCollectionOutput:
@@ -386,30 +384,26 @@ class ColorField(BaseModel):
return (self.r, self.g, self.b, self.a)
@invocation_output("color_output")
class ColorOutput(BaseInvocationOutput):
"""Base class for nodes that output a single color"""
type: Literal["color_output"] = "color_output"
color: ColorField = OutputField(description="The output color")
@invocation_output("color_collection_output")
class ColorCollectionOutput(BaseInvocationOutput):
"""Base class for nodes that output a collection of colors"""
type: Literal["color_collection_output"] = "color_collection_output"
# Outputs
collection: list[ColorField] = OutputField(description="The output colors", ui_type=UIType.ColorCollection)
collection: list[ColorField] = OutputField(
description="The output colors",
)
@title("Color Primitive")
@tags("primitives", "color")
@invocation("color", title="Color Primitive", tags=["primitives", "color"], category="primitives", version="1.0.0")
class ColorInvocation(BaseInvocation):
"""A color primitive value"""
type: Literal["color"] = "color"
# Inputs
color: ColorField = InputField(default=ColorField(r=0, g=0, b=0, a=255), description="The color value")
def invoke(self, context: InvocationContext) -> ColorOutput:
@@ -427,49 +421,51 @@ class ConditioningField(BaseModel):
conditioning_name: str = Field(description="The name of conditioning tensor")
@invocation_output("conditioning_output")
class ConditioningOutput(BaseInvocationOutput):
"""Base class for nodes that output a single conditioning tensor"""
type: Literal["conditioning_output"] = "conditioning_output"
conditioning: ConditioningField = OutputField(description=FieldDescriptions.cond)
@invocation_output("conditioning_collection_output")
class ConditioningCollectionOutput(BaseInvocationOutput):
"""Base class for nodes that output a collection of conditioning tensors"""
type: Literal["conditioning_collection_output"] = "conditioning_collection_output"
# Outputs
collection: list[ConditioningField] = OutputField(
description="The output conditioning tensors",
ui_type=UIType.ConditioningCollection,
)
@title("Conditioning Primitive")
@tags("primitives", "conditioning")
@invocation(
"conditioning",
title="Conditioning Primitive",
tags=["primitives", "conditioning"],
category="primitives",
version="1.0.0",
)
class ConditioningInvocation(BaseInvocation):
"""A conditioning tensor primitive value"""
type: Literal["conditioning"] = "conditioning"
conditioning: ConditioningField = InputField(description=FieldDescriptions.cond, input=Input.Connection)
def invoke(self, context: InvocationContext) -> ConditioningOutput:
return ConditioningOutput(conditioning=self.conditioning)
@title("Conditioning Primitive Collection")
@tags("primitives", "conditioning", "collection")
@invocation(
"conditioning_collection",
title="Conditioning Collection Primitive",
tags=["primitives", "conditioning", "collection"],
category="primitives",
version="1.0.0",
)
class ConditioningCollectionInvocation(BaseInvocation):
"""A collection of conditioning tensor primitive values"""
type: Literal["conditioning_collection"] = "conditioning_collection"
# Inputs
collection: list[ConditioningField] = InputField(
default=0, description="The collection of conditioning tensors", ui_type=UIType.ConditioningCollection
default_factory=list,
description="The collection of conditioning tensors",
)
def invoke(self, context: InvocationContext) -> ConditioningCollectionOutput:

18
invokeai/app/invocations/prompt.py
View File

@@ -1,5 +1,5 @@
from os.path import exists
from typing import Literal, Optional, Union
from typing import Optional, Union
import numpy as np
from dynamicprompts.generators import CombinatorialPromptGenerator, RandomPromptGenerator
@@ -7,17 +7,13 @@ from pydantic import validator
from invokeai.app.invocations.primitives import StringCollectionOutput
from .baseinvocation import BaseInvocation, InputField, InvocationContext, UIComponent, UIType, tags, title
from .baseinvocation import BaseInvocation, InputField, InvocationContext, UIComponent, invocation
@title("Dynamic Prompt")
@tags("prompt", "collection")
@invocation("dynamic_prompt", title="Dynamic Prompt", tags=["prompt", "collection"], category="prompt", version="1.0.0")
class DynamicPromptInvocation(BaseInvocation):
"""Parses a prompt using adieyal/dynamicprompts' random or combinatorial generator"""
type: Literal["dynamic_prompt"] = "dynamic_prompt"
# Inputs
prompt: str = InputField(description="The prompt to parse with dynamicprompts", ui_component=UIComponent.Textarea)
max_prompts: int = InputField(default=1, description="The number of prompts to generate")
combinatorial: bool = InputField(default=False, description="Whether to use the combinatorial generator")
@@ -33,15 +29,11 @@ class DynamicPromptInvocation(BaseInvocation):
return StringCollectionOutput(collection=prompts)
@title("Prompts from File")
@tags("prompt", "file")
@invocation("prompt_from_file", title="Prompts from File", tags=["prompt", "file"], category="prompt", version="1.0.0")
class PromptsFromFileInvocation(BaseInvocation):
"""Loads prompts from a text file"""
type: Literal["prompt_from_file"] = "prompt_from_file"
# Inputs
file_path: str = InputField(description="Path to prompt text file", ui_type=UIType.FilePath)
file_path: str = InputField(description="Path to prompt text file")
pre_prompt: Optional[str] = InputField(
default=None, description="String to prepend to each prompt", ui_component=UIComponent.Textarea
)

30
invokeai/app/invocations/sdxl.py
View File

@@ -1,5 +1,3 @@
from typing import Literal
from ...backend.model_management import ModelType, SubModelType
from .baseinvocation import (
BaseInvocation,
@@ -10,41 +8,35 @@ from .baseinvocation import (
InvocationContext,
OutputField,
UIType,
tags,
title,
invocation,
invocation_output,
)
from .model import ClipField, MainModelField, ModelInfo, UNetField, VaeField
@invocation_output("sdxl_model_loader_output")
class SDXLModelLoaderOutput(BaseInvocationOutput):
"""SDXL base model loader output"""
type: Literal["sdxl_model_loader_output"] = "sdxl_model_loader_output"
unet: UNetField = OutputField(description=FieldDescriptions.unet, title="UNet")
clip: ClipField = OutputField(description=FieldDescriptions.clip, title="CLIP 1")
clip2: ClipField = OutputField(description=FieldDescriptions.clip, title="CLIP 2")
vae: VaeField = OutputField(description=FieldDescriptions.vae, title="VAE")
@invocation_output("sdxl_refiner_model_loader_output")
class SDXLRefinerModelLoaderOutput(BaseInvocationOutput):
"""SDXL refiner model loader output"""
type: Literal["sdxl_refiner_model_loader_output"] = "sdxl_refiner_model_loader_output"
unet: UNetField = OutputField(description=FieldDescriptions.unet, title="UNet")
clip2: ClipField = OutputField(description=FieldDescriptions.clip, title="CLIP 2")
vae: VaeField = OutputField(description=FieldDescriptions.vae, title="VAE")
@title("SDXL Main Model")
@tags("model", "sdxl")
@invocation("sdxl_model_loader", title="SDXL Main Model", tags=["model", "sdxl"], category="model", version="1.0.0")
class SDXLModelLoaderInvocation(BaseInvocation):
"""Loads an sdxl base model, outputting its submodels."""
type: Literal["sdxl_model_loader"] = "sdxl_model_loader"
# Inputs
model: MainModelField = InputField(
description=FieldDescriptions.sdxl_main_model, input=Input.Direct, ui_type=UIType.SDXLMainModel
)
@@ -122,14 +114,16 @@ class SDXLModelLoaderInvocation(BaseInvocation):
)
@title("SDXL Refiner Model")
@tags("model", "sdxl", "refiner")
@invocation(
"sdxl_refiner_model_loader",
title="SDXL Refiner Model",
tags=["model", "sdxl", "refiner"],
category="model",
version="1.0.0",
)
class SDXLRefinerModelLoaderInvocation(BaseInvocation):
"""Loads an sdxl refiner model, outputting its submodels."""
type: Literal["sdxl_refiner_model_loader"] = "sdxl_refiner_model_loader"
# Inputs
model: MainModelField = InputField(
description=FieldDescriptions.sdxl_refiner_model,
input=Input.Direct,

9
invokeai/app/invocations/upscale.py
View File

@@ -11,7 +11,7 @@ from invokeai.app.invocations.primitives import ImageField, ImageOutput
from invokeai.app.models.image import ImageCategory, ResourceOrigin
from .baseinvocation import BaseInvocation, InputField, InvocationContext, title, tags
from .baseinvocation import BaseInvocation, InputField, InvocationContext, invocation
# TODO: Populate this from disk?
# TODO: Use model manager to load?
@@ -23,14 +23,10 @@ ESRGAN_MODELS = Literal[
]
@title("Upscale (RealESRGAN)")
@tags("esrgan", "upscale")
@invocation("esrgan", title="Upscale (RealESRGAN)", tags=["esrgan", "upscale"], category="esrgan", version="1.0.0")
class ESRGANInvocation(BaseInvocation):
"""Upscales an image using RealESRGAN."""
type: Literal["esrgan"] = "esrgan"
# Inputs
image: ImageField = InputField(description="The input image")
model_name: ESRGAN_MODELS = InputField(default="RealESRGAN_x4plus.pth", description="The Real-ESRGAN model to use")
@@ -110,6 +106,7 @@ class ESRGANInvocation(BaseInvocation):
node_id=self.id,
session_id=context.graph_execution_state_id,
is_intermediate=self.is_intermediate,
workflow=self.workflow,
)
return ImageOutput(

1
invokeai/app/services/config/__init__.py
View File

@@ -6,3 +6,4 @@ from .invokeai_config import ( # noqa F401
InvokeAIAppConfig,
get_invokeai_config,
)
from .base import PagingArgumentParser # noqa F401

34
invokeai/app/services/graph.py
View File

@@ -3,7 +3,7 @@
import copy
import itertools
import uuid
from typing import Annotated, Any, Literal, Optional, Union, get_args, get_origin, get_type_hints
from typing import Annotated, Any, Optional, Union, get_args, get_origin, get_type_hints
import networkx as nx
from pydantic import BaseModel, root_validator, validator
@@ -14,11 +14,13 @@ from ..invocations import * # noqa: F401 F403
from ..invocations.baseinvocation import (
BaseInvocation,
BaseInvocationOutput,
invocation,
Input,
InputField,
InvocationContext,
OutputField,
UIType,
invocation_output,
)
# in 3.10 this would be "from types import NoneType"
@@ -110,6 +112,10 @@ def are_connection_types_compatible(from_type: Any, to_type: Any) -> bool:
if to_type in get_args(from_type):
return True
# allow int -> float, pydantic will cast for us
if from_type is int and to_type is float:
return True
# if not issubclass(from_type, to_type):
if not is_union_subtype(from_type, to_type):
return False
@@ -148,24 +154,16 @@ class NodeAlreadyExecutedError(Exception):
# TODO: Create and use an Empty output?
@invocation_output("graph_output")
class GraphInvocationOutput(BaseInvocationOutput):
type: Literal["graph_output"] = "graph_output"
class Config:
schema_extra = {
"required": [
"type",
"image",
]
}
pass
# TODO: Fill this out and move to invocations
@invocation("graph")
class GraphInvocation(BaseInvocation):
"""Execute a graph"""
type: Literal["graph"] = "graph"
# TODO: figure out how to create a default here
graph: "Graph" = Field(description="The graph to run", default=None)
@@ -174,22 +172,20 @@ class GraphInvocation(BaseInvocation):
return GraphInvocationOutput()
@invocation_output("iterate_output")
class IterateInvocationOutput(BaseInvocationOutput):
"""Used to connect iteration outputs. Will be expanded to a specific output."""
type: Literal["iterate_output"] = "iterate_output"
item: Any = OutputField(
description="The item being iterated over", title="Collection Item", ui_type=UIType.CollectionItem
)
# TODO: Fill this out and move to invocations
@invocation("iterate")
class IterateInvocation(BaseInvocation):
"""Iterates over a list of items"""
type: Literal["iterate"] = "iterate"
collection: list[Any] = InputField(
description="The list of items to iterate over", default_factory=list, ui_type=UIType.Collection
)
@@ -200,19 +196,17 @@ class IterateInvocation(BaseInvocation):
return IterateInvocationOutput(item=self.collection[self.index])
@invocation_output("collect_output")
class CollectInvocationOutput(BaseInvocationOutput):
type: Literal["collect_output"] = "collect_output"
collection: list[Any] = OutputField(
description="The collection of input items", title="Collection", ui_type=UIType.Collection
)
@invocation("collect")
class CollectInvocation(BaseInvocation):
"""Collects values into a collection"""
type: Literal["collect"] = "collect"
item: Any = InputField(
description="The item to collect (all inputs must be of the same type)",
ui_type=UIType.CollectionItem,

23
invokeai/app/services/image_file_storage.py
View File

@@ -60,7 +60,7 @@ class ImageFileStorageBase(ABC):
image: PILImageType,
image_name: str,
metadata: Optional[dict] = None,
graph: Optional[dict] = None,
workflow: Optional[str] = None,
thumbnail_size: int = 256,
) -> None:
"""Saves an image and a 256x256 WEBP thumbnail. Returns a tuple of the image name, thumbnail name, and created timestamp."""
@@ -110,7 +110,7 @@ class DiskImageFileStorage(ImageFileStorageBase):
image: PILImageType,
image_name: str,
metadata: Optional[dict] = None,
graph: Optional[dict] = None,
workflow: Optional[str] = None,
thumbnail_size: int = 256,
) -> None:
try:
@@ -119,12 +119,23 @@ class DiskImageFileStorage(ImageFileStorageBase):
pnginfo = PngImagePlugin.PngInfo()
if metadata is not None:
pnginfo.add_text("invokeai_metadata", json.dumps(metadata))
if graph is not None:
pnginfo.add_text("invokeai_graph", json.dumps(graph))
if metadata is not None or workflow is not None:
if metadata is not None:
pnginfo.add_text("invokeai_metadata", json.dumps(metadata))
if workflow is not None:
pnginfo.add_text("invokeai_workflow", workflow)
else:
# For uploaded images, we want to retain metadata. PIL strips it on save; manually add it back
# TODO: retain non-invokeai metadata on save...
original_metadata = image.info.get("invokeai_metadata", None)
if original_metadata is not None:
pnginfo.add_text("invokeai_metadata", original_metadata)
original_workflow = image.info.get("invokeai_workflow", None)
if original_workflow is not None:
pnginfo.add_text("invokeai_workflow", original_workflow)
image.save(image_path, "PNG", pnginfo=pnginfo)
thumbnail_name = get_thumbnail_name(image_name)
thumbnail_path = self.get_path(thumbnail_name, thumbnail=True)
thumbnail_image = make_thumbnail(image, thumbnail_size)

24
invokeai/app/services/images.py
View File

@@ -54,6 +54,7 @@ class ImageServiceABC(ABC):
board_id: Optional[str] = None,
is_intermediate: bool = False,
metadata: Optional[dict] = None,
workflow: Optional[str] = None,
) -> ImageDTO:
"""Creates an image, storing the file and its metadata."""
pass
@@ -177,6 +178,7 @@ class ImageService(ImageServiceABC):
board_id: Optional[str] = None,
is_intermediate: bool = False,
metadata: Optional[dict] = None,
workflow: Optional[str] = None,
) -> ImageDTO:
if image_origin not in ResourceOrigin:
raise InvalidOriginException
@@ -186,16 +188,16 @@ class ImageService(ImageServiceABC):
image_name = self._services.names.create_image_name()
graph = None
if session_id is not None:
session_raw = self._services.graph_execution_manager.get_raw(session_id)
if session_raw is not None:
try:
graph = get_metadata_graph_from_raw_session(session_raw)
except Exception as e:
self._services.logger.warn(f"Failed to parse session graph: {e}")
graph = None
# TODO: Do we want to store the graph in the image at all? I don't think so...
# graph = None
# if session_id is not None:
# session_raw = self._services.graph_execution_manager.get_raw(session_id)
# if session_raw is not None:
# try:
# graph = get_metadata_graph_from_raw_session(session_raw)
# except Exception as e:
# self._services.logger.warn(f"Failed to parse session graph: {e}")
# graph = None
(width, height) = image.size
@@ -217,7 +219,7 @@ class ImageService(ImageServiceABC):
)
if board_id is not None:
self._services.board_image_records.add_image_to_board(board_id=board_id, image_name=image_name)
self._services.image_files.save(image_name=image_name, image=image, metadata=metadata, graph=graph)
self._services.image_files.save(image_name=image_name, image=image, metadata=metadata, workflow=workflow)
image_dto = self.get_dto(image_name)
return image_dto

2
invokeai/app/services/models/image_record.py
View File

@@ -53,7 +53,7 @@ class ImageRecordChanges(BaseModelExcludeNull, extra=Extra.forbid):
- `starred`: change whether the image is starred
"""
image_category: Optional[ImageCategory] = Field(description="The image's new category.")
image_category: Optional[ImageCategory] = Field(default=None, description="The image's new category.")
"""The image's new category."""
session_id: Optional[StrictStr] = Field(
default=None,