Split ip_adapter_conditioning out from ConditioningData.

invokeai/app/invocations/latent.py

@@ -488,7 +488,6 @@ class DenoiseLatentsInvocation(BaseInvocation):
         self,
         context: InvocationContext,
         ip_adapter: Optional[Union[IPAdapterField, list[IPAdapterField]]],
-        conditioning_data: ConditioningData,
         exit_stack: ExitStack,
     ) -> Optional[list[IPAdapterData]]:
         """If IP-Adapter is enabled, then this function loads the requisite models, and adds the image prompt embeddings
@@ -505,7 +504,6 @@ class DenoiseLatentsInvocation(BaseInvocation):
             return None
 
         ip_adapter_data_list = []
-        conditioning_data.ip_adapter_conditioning = []
         for single_ip_adapter in ip_adapter:
             ip_adapter_model: Union[IPAdapter, IPAdapterPlus] = exit_stack.enter_context(
                 context.models.load(single_ip_adapter.ip_adapter_model)
@@ -528,16 +526,13 @@ class DenoiseLatentsInvocation(BaseInvocation):
                     single_ipa_images, image_encoder_model
                 )
 
-                conditioning_data.ip_adapter_conditioning.append(
-                    IPAdapterConditioningInfo(image_prompt_embeds, uncond_image_prompt_embeds)
-                )
-
             ip_adapter_data_list.append(
                 IPAdapterData(
                     ip_adapter_model=ip_adapter_model,
                     weight=single_ip_adapter.weight,
                     begin_step_percent=single_ip_adapter.begin_step_percent,
                     end_step_percent=single_ip_adapter.end_step_percent,
+                    ip_adapter_conditioning=IPAdapterConditioningInfo(image_prompt_embeds, uncond_image_prompt_embeds),
                 )
             )
 
@@ -772,7 +767,6 @@ class DenoiseLatentsInvocation(BaseInvocation):
                 ip_adapter_data = self.prep_ip_adapter_data(
                     context=context,
                     ip_adapter=self.ip_adapter,
-                    conditioning_data=conditioning_data,
                     exit_stack=exit_stack,
                 )
 

invokeai/backend/stable_diffusion/diffusers_pipeline.py

@@ -23,7 +23,7 @@ from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
 from invokeai.app.services.config.config_default import get_config
 from invokeai.backend.ip_adapter.ip_adapter import IPAdapter
 from invokeai.backend.ip_adapter.unet_patcher import UNetPatcher
-from invokeai.backend.stable_diffusion.diffusion.conditioning_data import ConditioningData
+from invokeai.backend.stable_diffusion.diffusion.conditioning_data import ConditioningData, IPAdapterConditioningInfo
 from invokeai.backend.stable_diffusion.diffusion.shared_invokeai_diffusion import InvokeAIDiffuserComponent
 from invokeai.backend.util.attention import auto_detect_slice_size
 from invokeai.backend.util.devices import normalize_device
@@ -151,10 +151,11 @@ class ControlNetData:
 
 @dataclass
 class IPAdapterData:
-    ip_adapter_model: IPAdapter = Field(default=None)
+    ip_adapter_model: IPAdapter
-    # TODO: change to polymorphic so can do different weights per step (once implemented...)
+    ip_adapter_conditioning: IPAdapterConditioningInfo
+
+    # Either a single weight applied to all steps, or a list of weights for each step.
     weight: Union[float, List[float]] = Field(default=1.0)
-    # weight: float = Field(default=1.0)
     begin_step_percent: float = Field(default=0.0)
     end_step_percent: float = Field(default=1.0)
 
@@ -549,12 +550,17 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
 
             down_intrablock_additional_residuals = accum_adapter_state
 
+        ip_adapter_conditioning = None
+        if ip_adapter_data is not None:
+            ip_adapter_conditioning = [ipa.ip_adapter_conditioning for ipa in ip_adapter_data]
+
         uc_noise_pred, c_noise_pred = self.invokeai_diffuser.do_unet_step(
             sample=latent_model_input,
             timestep=t,  # TODO: debug how handled batched and non batched timesteps
             step_index=step_index,
             total_step_count=total_step_count,
             conditioning_data=conditioning_data,
+            ip_adapter_conditioning=ip_adapter_conditioning,
             down_block_additional_residuals=down_block_additional_residuals,  # for ControlNet
             mid_block_additional_residual=mid_block_additional_residual,  # for ControlNet
             down_intrablock_additional_residuals=down_intrablock_additional_residuals,  # for T2I-Adapter

invokeai/backend/stable_diffusion/diffusion/conditioning_data.py

@@ -69,5 +69,3 @@ class ConditioningData:
      ref [Common Diffusion Noise Schedules and Sample Steps are Flawed](https://arxiv.org/pdf/2305.08891.pdf)
     """
     guidance_rescale_multiplier: float = 0
-
-    ip_adapter_conditioning: Optional[list[IPAdapterConditioningInfo]] = None

invokeai/backend/stable_diffusion/diffusion/shared_invokeai_diffusion.py

@@ -12,6 +12,7 @@ from invokeai.app.services.config.config_default import get_config
 from invokeai.backend.stable_diffusion.diffusion.conditioning_data import (
     ConditioningData,
     ExtraConditioningInfo,
+    IPAdapterConditioningInfo,
     SDXLConditioningInfo,
 )
 
@@ -199,6 +200,7 @@ class InvokeAIDiffuserComponent:
         sample: torch.Tensor,
         timestep: torch.Tensor,
         conditioning_data: ConditioningData,
+        ip_adapter_conditioning: Optional[list[IPAdapterConditioningInfo]],
         step_index: int,
         total_step_count: int,
         down_block_additional_residuals: Optional[torch.Tensor] = None,  # for ControlNet
@@ -223,6 +225,7 @@ class InvokeAIDiffuserComponent:
                 x=sample,
                 sigma=timestep,
                 conditioning_data=conditioning_data,
+                ip_adapter_conditioning=ip_adapter_conditioning,
                 cross_attention_control_types_to_do=cross_attention_control_types_to_do,
                 down_block_additional_residuals=down_block_additional_residuals,
                 mid_block_additional_residual=mid_block_additional_residual,
@@ -236,6 +239,7 @@ class InvokeAIDiffuserComponent:
                 x=sample,
                 sigma=timestep,
                 conditioning_data=conditioning_data,
+                ip_adapter_conditioning=ip_adapter_conditioning,
                 down_block_additional_residuals=down_block_additional_residuals,
                 mid_block_additional_residual=mid_block_additional_residual,
                 down_intrablock_additional_residuals=down_intrablock_additional_residuals,
@@ -297,6 +301,7 @@ class InvokeAIDiffuserComponent:
         x,
         sigma,
         conditioning_data: ConditioningData,
+        ip_adapter_conditioning: Optional[list[IPAdapterConditioningInfo]],
         down_block_additional_residuals: Optional[torch.Tensor] = None,  # for ControlNet
         mid_block_additional_residual: Optional[torch.Tensor] = None,  # for ControlNet
         down_intrablock_additional_residuals: Optional[torch.Tensor] = None,  # for T2I-Adapter
@@ -308,14 +313,14 @@ class InvokeAIDiffuserComponent:
         sigma_twice = torch.cat([sigma] * 2)
 
         cross_attention_kwargs = None
-        if conditioning_data.ip_adapter_conditioning is not None:
+        if ip_adapter_conditioning is not None:
             # Note that we 'stack' to produce tensors of shape (batch_size, num_ip_images, seq_len, token_len).
             cross_attention_kwargs = {
                 "ip_adapter_image_prompt_embeds": [
                     torch.stack(
                         [ipa_conditioning.uncond_image_prompt_embeds, ipa_conditioning.cond_image_prompt_embeds]
                     )
-                    for ipa_conditioning in conditioning_data.ip_adapter_conditioning
+                    for ipa_conditioning in ip_adapter_conditioning
                 ]
             }
 
@@ -361,6 +366,7 @@ class InvokeAIDiffuserComponent:
         x: torch.Tensor,
         sigma,
         conditioning_data: ConditioningData,
+        ip_adapter_conditioning: Optional[list[IPAdapterConditioningInfo]],
         cross_attention_control_types_to_do: list[CrossAttentionType],
         down_block_additional_residuals: Optional[torch.Tensor] = None,  # for ControlNet
         mid_block_additional_residual: Optional[torch.Tensor] = None,  # for ControlNet
@@ -411,12 +417,12 @@ class InvokeAIDiffuserComponent:
         cross_attention_kwargs = None
 
         # Prepare IP-Adapter cross-attention kwargs for the unconditioned pass.
-        if conditioning_data.ip_adapter_conditioning is not None:
+        if ip_adapter_conditioning is not None:
             # Note that we 'unsqueeze' to produce tensors of shape (batch_size=1, num_ip_images, seq_len, token_len).
             cross_attention_kwargs = {
                 "ip_adapter_image_prompt_embeds": [
                     torch.unsqueeze(ipa_conditioning.uncond_image_prompt_embeds, dim=0)
-                    for ipa_conditioning in conditioning_data.ip_adapter_conditioning
+                    for ipa_conditioning in ip_adapter_conditioning
                 ]
             }
 
@@ -452,12 +458,12 @@ class InvokeAIDiffuserComponent:
         cross_attention_kwargs = None
 
         # Prepare IP-Adapter cross-attention kwargs for the conditioned pass.
-        if conditioning_data.ip_adapter_conditioning is not None:
+        if ip_adapter_conditioning is not None:
             # Note that we 'unsqueeze' to produce tensors of shape (batch_size=1, num_ip_images, seq_len, token_len).
             cross_attention_kwargs = {
                 "ip_adapter_image_prompt_embeds": [
                     torch.unsqueeze(ipa_conditioning.cond_image_prompt_embeds, dim=0)
-                    for ipa_conditioning in conditioning_data.ip_adapter_conditioning
+                    for ipa_conditioning in ip_adapter_conditioning
                 ]
             }