Merge branch 'main' into copilot/enhancement-allow-shared-boards

Author: JPPhoto

invokeai/app/invocations/canvas.py

@@ -0,0 +1,27 @@
+from invokeai.app.invocations.baseinvocation import BaseInvocation, invocation
+from invokeai.app.invocations.fields import FieldDescriptions, ImageField, InputField
+from invokeai.app.invocations.primitives import ImageOutput
+from invokeai.app.services.shared.invocation_context import InvocationContext
+
+
+@invocation(
+    "canvas_output",
+    title="Canvas Output",
+    tags=["canvas", "output", "image"],
+    category="canvas",
+    version="1.0.0",
+    use_cache=False,
+)
+class CanvasOutputInvocation(BaseInvocation):
+    """Outputs an image to the canvas staging area.
+
+    Use this node in workflows intended for canvas workflow integration.
+    Connect the final image of your workflow to this node to send it
+    to the canvas staging area when run via 'Run Workflow on Canvas'."""
+
+    image: ImageField = InputField(description=FieldDescriptions.image)
+
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        image = context.images.get_pil(self.image.image_name)
+        image_dto = context.images.save(image=image)
+        return ImageOutput.build(image_dto)

invokeai/backend/model_manager/configs/main.py

@@ -323,16 +323,26 @@ def _is_flux2_model(state_dict: dict[str | int, Any]) -> bool:
     return False
 
 
+def _filename_suggests_base(name: str) -> bool:
+    """Check if a model name/filename suggests it is a Base (undistilled) variant.
+
+    Klein 9B Base and Klein 9B have identical architectures and cannot be distinguished
+    from the state dict. We use the filename as a heuristic: filenames containing "base"
+    (e.g. "flux-2-klein-base-9b", "FLUX.2-klein-base-9B") indicate the undistilled model.
+    """
+    return "base" in name.lower()
+
+
 def _get_flux2_variant(state_dict: dict[str | int, Any]) -> Flux2VariantType | None:
     """Determine FLUX.2 variant from state dict.
 
     Distinguishes between Klein 4B and Klein 9B based on context embedding dimension:
     - Klein 4B: context_in_dim = 7680 (3 × Qwen3-4B hidden_size 2560)
     - Klein 9B: context_in_dim = 12288 (3 × Qwen3-8B hidden_size 4096)
 
-    Note: Klein 9B Base (undistilled) also has context_in_dim = 12288 but is rare.
-    We default to Klein9B (distilled) for all 9B models since GGUF models may not
-    include guidance embedding keys needed to distinguish them.
+    Note: Klein 9B (distilled) and Klein 9B Base (undistilled) have identical architectures
+    and cannot be distinguished from the state dict alone. This function defaults to Klein9B
+    for all 9B models. Callers should use filename heuristics to detect Klein9BBase.
 
     Supports both BFL format (checkpoint) and diffusers format keys:
     - BFL format: txt_in.weight (context embedder)
@@ -366,7 +376,7 @@ def _get_flux2_variant(state_dict: dict[str | int, Any]) -> Flux2VariantType | N
                 context_in_dim = shape[1]
                 # Determine variant based on context dimension
                 if context_in_dim == KLEIN_9B_CONTEXT_DIM:
-                    # Default to Klein9B (distilled) - the official/common 9B model
+                    # Default to Klein9B - callers use filename heuristics to detect Klein9BBase
                     return Flux2VariantType.Klein9B
                 elif context_in_dim == KLEIN_4B_CONTEXT_DIM:
                     return Flux2VariantType.Klein4B
@@ -553,6 +563,11 @@ def _get_variant_or_raise(cls, mod: ModelOnDisk) -> Flux2VariantType:
         if variant is None:
             raise NotAMatchError("unable to determine FLUX.2 model variant from state dict")
 
+        # Klein 9B Base and Klein 9B have identical architectures.
+        # Use filename heuristic to detect the Base (undistilled) variant.
+        if variant == Flux2VariantType.Klein9B and _filename_suggests_base(mod.name):
+            return Flux2VariantType.Klein9BBase
+
         return variant
 
     @classmethod
@@ -720,6 +735,11 @@ def _get_variant_or_raise(cls, mod: ModelOnDisk) -> Flux2VariantType:
         if variant is None:
             raise NotAMatchError("unable to determine FLUX.2 model variant from state dict")
 
+        # Klein 9B Base and Klein 9B have identical architectures.
+        # Use filename heuristic to detect the Base (undistilled) variant.
+        if variant == Flux2VariantType.Klein9B and _filename_suggests_base(mod.name):
+            return Flux2VariantType.Klein9BBase
+
         return variant
 
     @classmethod
@@ -829,30 +849,21 @@ def _get_variant_or_raise(cls, mod: ModelOnDisk) -> Flux2VariantType:
         - Klein 4B: joint_attention_dim = 7680 (3×Qwen3-4B hidden size)
         - Klein 9B/9B Base: joint_attention_dim = 12288 (3×Qwen3-8B hidden size)
 
-        To distinguish Klein 9B (distilled) from Klein 9B Base (undistilled),
-        we check guidance_embeds:
-        - Klein 9B (distilled): guidance_embeds = False (guidance is "baked in" during distillation)
-        - Klein 9B Base (undistilled): guidance_embeds = True (needs guidance at inference)
-
-        Note: The official BFL Klein 9B model is the distilled version with guidance_embeds=False.
+        Klein 9B (distilled) and Klein 9B Base (undistilled) have identical architectures
+        and both have guidance_embeds=False. We use a filename heuristic to detect Base models.
         """
         KLEIN_4B_CONTEXT_DIM = 7680  # 3 × 2560
         KLEIN_9B_CONTEXT_DIM = 12288  # 3 × 4096
 
         transformer_config = get_config_dict_or_raise(mod.path / "transformer" / "config.json")
 
         joint_attention_dim = transformer_config.get("joint_attention_dim", 4096)
-        guidance_embeds = transformer_config.get("guidance_embeds", False)
 
         # Determine variant based on joint_attention_dim
         if joint_attention_dim == KLEIN_9B_CONTEXT_DIM:
-            # Check guidance_embeds to distinguish distilled from undistilled
-            # Klein 9B (distilled): guidance_embeds = False (guidance is baked in)
-            # Klein 9B Base (undistilled): guidance_embeds = True (needs guidance)
-            if guidance_embeds:
+            if _filename_suggests_base(mod.name):
                 return Flux2VariantType.Klein9BBase
-            else:
-                return Flux2VariantType.Klein9B
+            return Flux2VariantType.Klein9B
         elif joint_attention_dim == KLEIN_4B_CONTEXT_DIM:
             return Flux2VariantType.Klein4B
         elif joint_attention_dim > 4096:

invokeai/backend/model_manager/starter_models.py

@@ -71,7 +71,7 @@ class StarterModelBundle(BaseModel):
     name="t5_base_encoder",
     base=BaseModelType.Any,
     source="InvokeAI/t5-v1_1-xxl::bfloat16",
-    description="T5-XXL text encoder (used in FLUX pipelines). ~8GB",
+    description="T5-XXL text encoder (used in FLUX pipelines). ~9.5GB",
     type=ModelType.T5Encoder,
 )
 
@@ -156,39 +156,39 @@ class StarterModelBundle(BaseModel):
     name="FLUX.1 Kontext dev (quantized)",
     base=BaseModelType.Flux,
     source="https://huggingface.co/unsloth/FLUX.1-Kontext-dev-GGUF/resolve/main/flux1-kontext-dev-Q4_K_M.gguf",
-    description="FLUX.1 Kontext dev quantized (q4_k_m). Total size with dependencies: ~14GB",
+    description="FLUX.1 Kontext dev quantized (q4_k_m). Total size with dependencies: ~12GB",
     type=ModelType.Main,
     dependencies=[t5_8b_quantized_encoder, flux_vae, clip_l_encoder],
 )
 flux_krea = StarterModel(
     name="FLUX.1 Krea dev",
     base=BaseModelType.Flux,
     source="https://huggingface.co/InvokeAI/FLUX.1-Krea-dev/resolve/main/flux1-krea-dev.safetensors",
-    description="FLUX.1 Krea dev. Total size with dependencies: ~33GB",
+    description="FLUX.1 Krea dev. Total size with dependencies: ~29GB",
     type=ModelType.Main,
     dependencies=[t5_8b_quantized_encoder, flux_vae, clip_l_encoder],
 )
 flux_krea_quantized = StarterModel(
     name="FLUX.1 Krea dev (quantized)",
     base=BaseModelType.Flux,
     source="https://huggingface.co/InvokeAI/FLUX.1-Krea-dev-GGUF/resolve/main/flux1-krea-dev-Q4_K_M.gguf",
-    description="FLUX.1 Krea dev quantized (q4_k_m). Total size with dependencies: ~14GB",
+    description="FLUX.1 Krea dev quantized (q4_k_m). Total size with dependencies: ~12GB",
     type=ModelType.Main,
     dependencies=[t5_8b_quantized_encoder, flux_vae, clip_l_encoder],
 )
 sd35_medium = StarterModel(
     name="SD3.5 Medium",
     base=BaseModelType.StableDiffusion3,
     source="stabilityai/stable-diffusion-3.5-medium",
-    description="Medium SD3.5 Model: ~15GB",
+    description="Medium SD3.5 Model: ~16GB",
     type=ModelType.Main,
     dependencies=[],
 )
 sd35_large = StarterModel(
     name="SD3.5 Large",
     base=BaseModelType.StableDiffusion3,
     source="stabilityai/stable-diffusion-3.5-large",
-    description="Large SD3.5 Model: ~19G",
+    description="Large SD3.5 Model: ~28GB",
     type=ModelType.Main,
     dependencies=[],
 )
@@ -644,7 +644,7 @@ class StarterModelBundle(BaseModel):
     name="CogView4",
     base=BaseModelType.CogView4,
     source="THUDM/CogView4-6B",
-    description="The base CogView4 model (~29GB).",
+    description="The base CogView4 model (~31GB).",
     type=ModelType.Main,
 )
 # endregion
@@ -695,7 +695,7 @@ class StarterModelBundle(BaseModel):
     name="FLUX.2 VAE",
     base=BaseModelType.Flux2,
     source="black-forest-labs/FLUX.2-klein-4B::vae",
-    description="FLUX.2 VAE (16-channel, same architecture as FLUX.1 VAE). ~335MB",
+    description="FLUX.2 VAE (16-channel, same architecture as FLUX.1 VAE). ~168MB",
     type=ModelType.VAE,
 )
 
@@ -719,7 +719,7 @@ class StarterModelBundle(BaseModel):
     name="FLUX.2 Klein 4B (Diffusers)",
     base=BaseModelType.Flux2,
     source="black-forest-labs/FLUX.2-klein-4B",
-    description="FLUX.2 Klein 4B in Diffusers format - includes transformer, VAE and Qwen3 encoder. ~10GB",
+    description="FLUX.2 Klein 4B in Diffusers format - includes transformer, VAE and Qwen3 encoder. ~16GB",
     type=ModelType.Main,
 )
 
@@ -745,7 +745,7 @@ class StarterModelBundle(BaseModel):
     name="FLUX.2 Klein 9B (Diffusers)",
     base=BaseModelType.Flux2,
     source="black-forest-labs/FLUX.2-klein-9B",
-    description="FLUX.2 Klein 9B in Diffusers format - includes transformer, VAE and Qwen3 encoder. ~20GB",
+    description="FLUX.2 Klein 9B in Diffusers format - includes transformer, VAE and Qwen3 encoder. ~35GB",
     type=ModelType.Main,
 )
 
@@ -821,7 +821,7 @@ class StarterModelBundle(BaseModel):
     name="Z-Image Turbo",
     base=BaseModelType.ZImage,
     source="Tongyi-MAI/Z-Image-Turbo",
-    description="Z-Image Turbo - fast 6B parameter text-to-image model with 8 inference steps. Supports bilingual prompts (English & Chinese). ~30.6GB",
+    description="Z-Image Turbo - fast 6B parameter text-to-image model with 8 inference steps. Supports bilingual prompts (English & Chinese). ~33GB",
     type=ModelType.Main,
 )
 

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

@@ -2391,6 +2391,27 @@
         "pullBboxIntoReferenceImageError": "Problem Pulling BBox Into ReferenceImage",
         "addAdjustments": "Add Adjustments",
         "removeAdjustments": "Remove Adjustments",
+        "workflowIntegration": {
+            "title": "Run Workflow on Canvas",
+            "description": "Select a workflow with a Canvas Output node and an image parameter to run on the current canvas layer. You can adjust parameters before executing. The result will be added back to the canvas.",
+            "execute": "Execute Workflow",
+            "executing": "Executing...",
+            "runWorkflow": "Run Workflow",
+            "filteringWorkflows": "Filtering workflows...",
+            "loadingWorkflows": "Loading workflows...",
+            "noWorkflowsFound": "No workflows found.",
+            "noWorkflowsWithImageField": "No compatible workflows found. A workflow needs a Form Builder with an image input field and a Canvas Output node.",
+            "selectWorkflow": "Select Workflow",
+            "selectPlaceholder": "Choose a workflow...",
+            "unnamedWorkflow": "Unnamed Workflow",
+            "loadingParameters": "Loading workflow parameters...",
+            "noFormBuilderError": "This workflow has no form builder and cannot be used. Please select a different workflow.",
+            "imageFieldSelected": "This field will receive the canvas image",
+            "imageFieldNotSelected": "Click to use this field for canvas image",
+            "executionStarted": "Workflow execution started",
+            "executionStartedDescription": "The result will appear in the staging area when complete.",
+            "executionFailed": "Failed to execute workflow"
+        },
         "compositeOperation": {
             "label": "Blend Mode",
             "add": "Add Blend Mode",

invokeai/frontend/web/src/app/components/GlobalModalIsolator.tsx

@@ -1,6 +1,7 @@
 import { GlobalImageHotkeys } from 'app/components/GlobalImageHotkeys';
 import ChangeBoardModal from 'features/changeBoardModal/components/ChangeBoardModal';
 import { CanvasPasteModal } from 'features/controlLayers/components/CanvasPasteModal';
+import { CanvasWorkflowIntegrationModal } from 'features/controlLayers/components/CanvasWorkflowIntegration/CanvasWorkflowIntegrationModal';
 import { CanvasManagerProviderGate } from 'features/controlLayers/contexts/CanvasManagerProviderGate';
 import { CropImageModal } from 'features/cropper/components/CropImageModal';
 import { DeleteImageModal } from 'features/deleteImageModal/components/DeleteImageModal';
@@ -51,6 +52,7 @@ export const GlobalModalIsolator = memo(() => {
       <SaveWorkflowAsDialog />
       <CanvasManagerProviderGate>
         <CanvasPasteModal />
+        <CanvasWorkflowIntegrationModal />
       </CanvasManagerProviderGate>
       <LoadWorkflowFromGraphModal />
       <CropImageModal />

invokeai/frontend/web/src/app/logging/logger.ts

@@ -16,6 +16,7 @@ const $logger = atom<Logger>(Roarr.child(BASE_CONTEXT));
 
 export const zLogNamespace = z.enum([
   'canvas',
+  'canvas-workflow-integration',
   'config',
   'dnd',
   'events',

invokeai/frontend/web/src/app/store/store.ts

@@ -25,6 +25,7 @@ import { canvasSettingsSliceConfig } from 'features/controlLayers/store/canvasSe
 import { canvasSliceConfig } from 'features/controlLayers/store/canvasSlice';
 import { canvasSessionSliceConfig } from 'features/controlLayers/store/canvasStagingAreaSlice';
 import { canvasTextSliceConfig } from 'features/controlLayers/store/canvasTextSlice';
+import { canvasWorkflowIntegrationSliceConfig } from 'features/controlLayers/store/canvasWorkflowIntegrationSlice';
 import { lorasSliceConfig } from 'features/controlLayers/store/lorasSlice';
 import { paramsSliceConfig } from 'features/controlLayers/store/paramsSlice';
 import { refImagesSliceConfig } from 'features/controlLayers/store/refImagesSlice';
@@ -67,6 +68,7 @@ const SLICE_CONFIGS = {
   [canvasSettingsSliceConfig.slice.reducerPath]: canvasSettingsSliceConfig,
   [canvasTextSliceConfig.slice.reducerPath]: canvasTextSliceConfig,
   [canvasSliceConfig.slice.reducerPath]: canvasSliceConfig,
+  [canvasWorkflowIntegrationSliceConfig.slice.reducerPath]: canvasWorkflowIntegrationSliceConfig,
   [changeBoardModalSliceConfig.slice.reducerPath]: changeBoardModalSliceConfig,
   [dynamicPromptsSliceConfig.slice.reducerPath]: dynamicPromptsSliceConfig,
   [gallerySliceConfig.slice.reducerPath]: gallerySliceConfig,
@@ -98,6 +100,7 @@ const ALL_REDUCERS = {
     canvasSliceConfig.slice.reducer,
     canvasSliceConfig.undoableConfig?.reduxUndoOptions
   ),
+  [canvasWorkflowIntegrationSliceConfig.slice.reducerPath]: canvasWorkflowIntegrationSliceConfig.slice.reducer,
   [changeBoardModalSliceConfig.slice.reducerPath]: changeBoardModalSliceConfig.slice.reducer,
   [dynamicPromptsSliceConfig.slice.reducerPath]: dynamicPromptsSliceConfig.slice.reducer,
   [gallerySliceConfig.slice.reducerPath]: gallerySliceConfig.slice.reducer,

invokeai/frontend/web/src/features/controlLayers/components/CanvasWorkflowIntegration/CanvasWorkflowIntegrationModal.tsx

@@ -0,0 +1,93 @@
+import {
+  Button,
+  ButtonGroup,
+  Flex,
+  Heading,
+  Modal,
+  ModalBody,
+  ModalCloseButton,
+  ModalContent,
+  ModalFooter,
+  ModalHeader,
+  ModalOverlay,
+  Spacer,
+  Spinner,
+  Text,
+} from '@invoke-ai/ui-library';
+import { useAppDispatch, useAppSelector } from 'app/store/storeHooks';
+import {
+  canvasWorkflowIntegrationClosed,
+  selectCanvasWorkflowIntegrationIsOpen,
+  selectCanvasWorkflowIntegrationIsProcessing,
+  selectCanvasWorkflowIntegrationSelectedWorkflowId,
+} from 'features/controlLayers/store/canvasWorkflowIntegrationSlice';
+import { memo, useCallback } from 'react';
+import { useTranslation } from 'react-i18next';
+
+import { CanvasWorkflowIntegrationParameterPanel } from './CanvasWorkflowIntegrationParameterPanel';
+import { CanvasWorkflowIntegrationWorkflowSelector } from './CanvasWorkflowIntegrationWorkflowSelector';
+import { useCanvasWorkflowIntegrationExecute } from './useCanvasWorkflowIntegrationExecute';
+
+export const CanvasWorkflowIntegrationModal = memo(() => {
+  const { t } = useTranslation();
+  const dispatch = useAppDispatch();
+
+  const isOpen = useAppSelector(selectCanvasWorkflowIntegrationIsOpen);
+  const isProcessing = useAppSelector(selectCanvasWorkflowIntegrationIsProcessing);
+  const selectedWorkflowId = useAppSelector(selectCanvasWorkflowIntegrationSelectedWorkflowId);
+
+  const { execute, canExecute } = useCanvasWorkflowIntegrationExecute();
+
+  const onClose = useCallback(() => {
+    if (!isProcessing) {
+      dispatch(canvasWorkflowIntegrationClosed());
+    }
+  }, [dispatch, isProcessing]);
+
+  const onExecute = useCallback(() => {
+    execute();
+  }, [execute]);
+
+  return (
+    <Modal isOpen={isOpen} onClose={onClose} size="xl" isCentered>
+      <ModalOverlay />
+      <ModalContent>
+        <ModalHeader>
+          <Heading size="md">{t('controlLayers.workflowIntegration.title')}</Heading>
+        </ModalHeader>
+        <ModalCloseButton isDisabled={isProcessing} />
+
+        <ModalBody>
+          <Flex direction="column" gap={4}>
+            <Text fontSize="sm" color="base.400">
+              {t('controlLayers.workflowIntegration.description')}
+            </Text>
+
+            <CanvasWorkflowIntegrationWorkflowSelector />
+
+            {selectedWorkflowId && <CanvasWorkflowIntegrationParameterPanel />}
+          </Flex>
+        </ModalBody>
+
+        <ModalFooter>
+          <ButtonGroup>
+            <Button variant="ghost" onClick={onClose} isDisabled={isProcessing}>
+              {t('common.cancel')}
+            </Button>
+            <Spacer />
+            <Button
+              onClick={onExecute}
+              isDisabled={!canExecute || isProcessing}
+              loadingText={t('controlLayers.workflowIntegration.executing')}
+            >
+              {isProcessing && <Spinner size="sm" mr={2} />}
+              {t('controlLayers.workflowIntegration.execute')}
+            </Button>
+          </ButtonGroup>
+        </ModalFooter>
+      </ModalContent>
+    </Modal>
+  );
+});
+
+CanvasWorkflowIntegrationModal.displayName = 'CanvasWorkflowIntegrationModal';