Update pipeline_stable_diffusion_3_ipa.py

pipeline_stable_diffusion_3_ipa.py

@@ -1196,14 +1196,17 @@ class StableDiffusion3Pipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSingle
             clip_image_embeds_5 = clip_image_embeds_5 * scale_5
             image_prompt_embeds_list.append(clip_image_embeds_5)
             
-        clip_image_embeds_cat_list = torch.cat(image_prompt_embeds_list).mean(dim=0)
-        print('catted embeds list with mean and unsqueeze: ',clip_image_embeds_cat_list.shape)
-        seq_len, _ = clip_image_embeds_cat_list.shape
-        clip_image_embeds_cat_list_repeat = clip_image_embeds_cat_list.repeat(1, 1, 1)
+        #clip_image_embeds_cat_list = torch.cat(image_prompt_embeds_list).mean(dim=0)
+        clip_image_embeds_cat_list = torch.cat(image_prompt_embeds_list)
+        print('catted embeds list with mean: ',clip_image_embeds_cat_list.shape)
+        #seq_len, _ = clip_image_embeds_cat_list.shape
+        #clip_image_embeds_cat_list_repeat = clip_image_embeds_cat_list.repeat(1, 1, 1)
+        clip_image_embeds_view = clip_image_embeds_cat.unsqueeze(0) # Added unsqueeze here instead
         print('catted embeds repeat: ',clip_image_embeds_cat_list_repeat.shape)
         clip_image_embeds_view = clip_image_embeds_cat_list_repeat.view(1, seq_len, -1)
         print('catted viewed: ',clip_image_embeds_view.shape)
-        zeros_tensor = torch.zeros_like(clip_image_embeds_view)
+        #zeros_tensor = torch.zeros_like(clip_image_embeds_view)
+        zeros_tensor = torch.zeros_like(clip_image_embeds_view[:, :image_prompt_embeds_list[0].shape[1], :]) # Make zeros tensor match the sequence length of a single image embedding
         print('zeros: ',zeros_tensor.shape)
         clip_image_embeds = torch.cat([zeros_tensor, clip_image_embeds_view], dim=0)
         print('embeds shape: ', clip_image_embeds.shape)