Update aduc_framework/managers/pipeline_wan_i2v (4).py

aduc_framework/managers/pipeline_wan_i2v (4).py

@@ -432,7 +432,7 @@ class WanImageToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
             latents = latents.to(device=device, dtype=dtype)
         print(f"latents{latents.shape}")
       
-        image = image.unsqueeze(2)  # [batch_size, channels, 1, height, width]
+        pipeline_wan_i2v = image.unsqueeze(2)  # [batch_size, channels, 1, height, width]
 
         if self.config.expand_timesteps:
             video_condition = image
@@ -651,6 +651,10 @@ class WanImageToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
         if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
             callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs
 
+
+
+        print(f"latents00{latents.shape}")
+      
         # 1. Check inputs. Raise error if not correct
         self.check_inputs(
             prompt,
@@ -719,6 +723,9 @@ class WanImageToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
             image_embeds = image_embeds.repeat(batch_size, 1, 1)
             image_embeds = image_embeds.to(transformer_dtype)
 
+
+        print(f"image_embeds{image_embeds.shape}")
+        
         # 4. Prepare timesteps
         self.scheduler.set_timesteps(num_inference_steps, device=device)
         timesteps = self.scheduler.timesteps
@@ -744,12 +751,21 @@ class WanImageToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
             latents,
             last_image,
         )
+
+
+        print(f"latents_outputs{latents_outputs.shape}")
+        
         if self.config.expand_timesteps:
             # wan 2.2 5b i2v use firt_frame_mask to mask timesteps
             latents, condition, first_frame_mask = latents_outputs
         else:
             latents, condition = latents_outputs
 
+
+
+        print(f"latentsxx{latents.shape}")
+            
+
         # 6. Denoising loop
         num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
         self._num_timesteps = len(timesteps)
@@ -812,6 +828,8 @@ class WanImageToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
                 # compute the previous noisy sample x_t -> x_t-1
                 latents = self.scheduler.step(noise_pred, t, latents, return_dict=False)[0]
 
+                print(f"latentsppp{latents.shape}")
+               
                 if callback_on_step_end is not None:
                     callback_kwargs = {}
                     for k in callback_on_step_end_tensor_inputs:
@@ -831,6 +849,10 @@ class WanImageToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
 
         self._current_timestep = None
 
+
+        print(f"latentsfim{latents.shape}")
+
+        
         if self.config.expand_timesteps:
             latents = (1 - first_frame_mask) * condition + first_frame_mask * latents