feat: Add ability to upload more poses and combine them with faces + error counting

app.py

@@ -15,7 +15,7 @@ AWS_ACCESS_SECRET = os.getenv('AWS_ACCESS_SECRET')
 
 def process_images(
     person_images,
-    pose_image,
+    pose_images,
     prompt,
     negative_prompt,
     num_steps,
@@ -35,7 +35,7 @@ def process_images(
         gr.Warning('No person images set')
         return
     
-    if pose_image == None:
+    if pose_images == None:
         gr.Warning('No pose image set')
         return
     
@@ -43,50 +43,94 @@ def process_images(
         gr.Warning('No prompt set')
         return
 
-    person_images_paths = [image for image in person_images]
-    pose_image_path = pose_image
-    image_paths = person_images_paths + [pose_image_path]
+    person_images_paths = person_images
+    pose_images_paths = pose_images
 
     yield [], "Uploading images"
 
-    uploaded_person_image_urls, uploaded_pose_image_urls = upload_images_concurrently(person_images_paths, [pose_image_path])
-    uploaded_pose_image_url = uploaded_pose_image_urls[0] if uploaded_pose_image_urls else None
+    uploaded_person_image_urls, uploaded_pose_image_urls = upload_images_concurrently(
+        person_images_paths=person_images_paths,
+        pose_images_paths=pose_images_paths,
+    )
 
-    requests_data = []
-    for i in range(0, generations_repeat_count):
-        for person_image_url in uploaded_person_image_urls:
-            requests_data.append(
-                {
-                    "faceImageUrl": person_image_url,
-                    "poseImageUrl": uploaded_pose_image_url,
-                    "prompt": prompt,
-                    "n_prompt": negative_prompt,
-                    "num_steps": num_steps,
-                    "identity_strength_ration": identity_strength_ration,
-                    "adapter_strength_ration": adapter_strength_ration,
-                    "pose_strength_ration": pose_strength_ration,
-                    "canny_strength_ration": canny_strength_ration,
-                    "depth_strength_ration": depth_strength_ration,
-                    "controlnet_selection": controlnet_selection,
-                    "guidance_strength_ration": guidance_strength_ration,
-                    "scheduler": scheduler,
-                    "enable_lcm": enable_lcm,
-                    "enhance_face_region": enhance_face_region
-                }
-            )
+    requests_data = generate_requests_data(
+        generations_repeat_count=generations_repeat_count,
+        uploaded_person_image_urls=uploaded_person_image_urls,
+        uploaded_pose_image_urls=uploaded_pose_image_urls,
+        prompt=prompt,
+        negative_prompt=negative_prompt,
+        num_steps=num_steps,
+        identity_strength_ration=identity_strength_ration,
+        adapter_strength_ration=adapter_strength_ration,
+        pose_strength_ration=pose_strength_ration,
+        canny_strength_ration=canny_strength_ration,
+        depth_strength_ration=depth_strength_ration,
+        controlnet_selection=controlnet_selection,
+        guidance_strength_ration=guidance_strength_ration,
+        scheduler=scheduler,
+        enable_lcm=enable_lcm,
+        enhance_face_region=enhance_face_region,
+    )
 
     yield [], f"Generating images 0/{len(requests_data)}"
 
     gallery_items = []
+    error_count = 0
     with concurrent.futures.ThreadPoolExecutor() as executor:
         futures = {executor.submit(execute_instantid_request, req_data.copy()): req_data for req_data in requests_data}
         for future in concurrent.futures.as_completed(futures):
             response = future.result()
             if response is not None:
                 gallery_items.append((response, "Caption"))
-                yield gallery_items, f"Generating images {len(gallery_items)}/{len(requests_data)}"
-    
-def upload_images_concurrently(person_images_paths, pose_images_paths, progress_callback):
+            else:
+                error_count += 1
+
+            yield gallery_items, f"Generating images {len(gallery_items)}/{len(requests_data)} (Failed: {error_count})"
+ 
+def generate_requests_data(
+    generations_repeat_count,
+    uploaded_person_image_urls,
+    uploaded_pose_image_urls,
+    prompt,
+    negative_prompt,
+    num_steps,
+    identity_strength_ration,
+    adapter_strength_ration,
+    pose_strength_ration,
+    canny_strength_ration,
+    depth_strength_ration,
+    controlnet_selection,
+    guidance_strength_ration,
+    scheduler,
+    enable_lcm,
+    enhance_face_region,
+):
+    requests_data = []
+    for i in range(0, generations_repeat_count):
+        for person_image_url in uploaded_person_image_urls:
+            for pose_image_url in uploaded_pose_image_urls:
+                requests_data.append(
+                    {
+                        "faceImageUrl": person_image_url,
+                        "poseImageUrl": pose_image_url,
+                        "prompt": prompt,
+                        "n_prompt": negative_prompt,
+                        "num_steps": num_steps,
+                        "identity_strength_ration": identity_strength_ration,
+                        "adapter_strength_ration": adapter_strength_ration,
+                        "pose_strength_ration": pose_strength_ration,
+                        "canny_strength_ration": canny_strength_ration,
+                        "depth_strength_ration": depth_strength_ration,
+                        "controlnet_selection": controlnet_selection,
+                        "guidance_strength_ration": guidance_strength_ration,
+                        "scheduler": scheduler,
+                        "enable_lcm": enable_lcm,
+                        "enhance_face_region": enhance_face_region
+                    }
+                )
+    return requests_data 
+
+def upload_images_concurrently(person_images_paths, pose_images_paths):
     """
     Uploads person and pose images concurrently and keeps them organized.
     Returns a tuple of lists: (list of person image URLs, list of pose image URLs).
@@ -167,9 +211,9 @@ with gr.Blocks() as demo:
             file_types=["jpg", "jpeg", "png", "webp"],
         )
 
-        pose_image = gr.File(
+        pose_images = gr.File(
             label="Pose",
-            file_count="single",
+            file_count="multiple",
             file_types=["jpg", "jpeg", "png", "webp"],
         )
 
@@ -296,7 +340,7 @@ with gr.Blocks() as demo:
         fn=process_images,
         inputs=[
             person_images,
-            pose_image,
+            pose_images,
             prompt,
             negative_prompt,
             num_steps,