Fix Stable Diffusion 3.x pooled prompt embedding with multiple images (#12306)

src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet.py

@@ -355,7 +355,7 @@ class StableDiffusion3ControlNetPipeline(
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet_inpainting.py

@@ -373,7 +373,7 @@ class StableDiffusion3ControlNetInpaintingPipeline(
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/pag/pipeline_pag_sd_3.py

@@ -326,7 +326,7 @@ class StableDiffusion3PAGPipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSin
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/pag/pipeline_pag_sd_3_img2img.py

@@ -342,7 +342,7 @@ class StableDiffusion3PAGImg2ImgPipeline(DiffusionPipeline, SD3LoraLoaderMixin,
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3.py

@@ -336,7 +336,7 @@ class StableDiffusion3Pipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSingle
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_img2img.py

@@ -361,7 +361,7 @@ class StableDiffusion3Img2ImgPipeline(DiffusionPipeline, SD3LoraLoaderMixin, Fro
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_inpaint.py

@@ -367,7 +367,7 @@ class StableDiffusion3InpaintPipeline(DiffusionPipeline, SD3LoraLoaderMixin, Fro
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds