Fixed use of ControlNet control_weight parameter

invokeai/app/invocations/controlnet_image_processors.py

@@ -35,8 +35,8 @@ class ControlField(BaseModel):
     # width: Optional[int] = Field(default=None, description="The width of the image in pixels")
     # height: Optional[int] = Field(default=None, description="The height of the image in pixels")
     # mode: Optional[str] = Field(default=None, description="The mode of the image")
-    control_model: Optional[str] = Field(default=None, description="The control model used")
-    control_weight: Optional[float] = Field(default=None, description="The control weight used")
+    control_model: Optional[str] = Field(default=None, description="control model used")
+    control_weight: Optional[float] = Field(default=None, description="weight given to controlnet")
 
     class Config:
         schema_extra = {
@@ -62,7 +62,7 @@ class ControlNetInvocation(BaseInvocation):
     # Inputs
     image: ImageField = Field(default=None, description="image to process")
     control_model: str = Field(default=None, description="control model to use")
-    control_weight: float = Field(default=0.5, ge=0, le=1, description="control weight")
+    control_weight: float = Field(default=0.5, ge=0, le=1, description="weight given to controlnet")
     # TODO: support additional ControlNet parameters (mostly just passthroughs to other nodes with ControlField inputs)
     # begin_step_percent: float = Field(default=0, ge=0, le=1,
     #                                    description="% of total steps at which controlnet is first applied")
@@ -78,7 +78,7 @@ class ControlNetInvocation(BaseInvocation):
             control=ControlField(
                 image=self.image,
                 control_model=self.control_model,
-                control_weight=self.control_weight,
+                control_weight=self.control_weight
             ),
         )
 

invokeai/app/invocations/latent.py

@@ -350,6 +350,7 @@ class TextToLatentsInvocation(BaseInvocation):
             conditioning_data=conditioning_data,
             callback=step_callback,
             control_image=control_images,
+            control_weight=control_weights,
         )
 
         # https://discuss.huggingface.co/t/memory-usage-by-later-pipeline-stages/23699

invokeai/backend/stable_diffusion/diffusers_pipeline.py

@@ -647,11 +647,11 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
 
         if (self.control_model is not None) and (kwargs.get("control_image") is not None):
             control_image = kwargs.get("control_image") # should be a processed tensor derived from the control image(s)
-            control_scale = kwargs.get("control_scale", 1.0)  # control_scale default is 1.0
-            # handling case where using multiple control models but only specifying single control_scale
-            #     so reshape control_scale to match number of control models
-            if isinstance(self.control_model, MultiControlNetModel) and isinstance(control_scale, float):
-                control_scale = [control_scale] * len(self.control_model.nets)
+            control_weight = kwargs.get("control_weight", 1.0)  # control_weight default is 1.0
+            # handling case where using multiple control models but only specifying single control_weight
+            #     so reshape control_weight to match number of control models
+            if isinstance(self.control_model, MultiControlNetModel) and isinstance(control_weight, float):
+                control_weight = [control_weight] * len(self.control_model.nets)
             if conditioning_data.guidance_scale > 1.0:
                 # expand the latents input to control model if doing classifier free guidance
                 #    (which I think for now is always true, there is conditional elsewhere that stops execution if
@@ -660,13 +660,15 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
             else:
                 latent_control_input = latent_model_input
             # controlnet inference
+
+            print("control_weight: ", control_weight)
             down_block_res_samples, mid_block_res_sample = self.control_model(
                 latent_control_input,
                 timestep,
                 encoder_hidden_states=torch.cat([conditioning_data.unconditioned_embeddings,
                                                  conditioning_data.text_embeddings]),
                 controlnet_cond=control_image,
-                conditioning_scale=control_scale,
+                conditioning_scale=control_weight,
                 return_dict=False,
             )
         else:

scripts/controlnet_legacy_txt2img_example.py

@@ -41,7 +41,7 @@ print("testing Txt2Img.generate() with control_image arg")
 outputs = txt2img_canny.generate(
     prompt="old man",
     control_image=canny_image,
-    control_scale=1.0,
+    control_weight=1.0,
     seed=0,
     num_steps=30,
     precision="float16",