pil

app.py

@@ -256,11 +256,11 @@ class PicletGeneratorService:
     FLUX_SPACE = "black-forest-labs/FLUX.1-schnell"
 
     @staticmethod
-    async def generate_enhanced_caption(image_data, hf_token: str) -> str:
+    async def generate_enhanced_caption(image_pil, hf_token: str) -> str:
         """Generate detailed image description using JoyCaption
 
         Args:
-            image_data: PIL Image or file path
+            image_pil: PIL Image object
             hf_token: User's HuggingFace token
         """
         try:
@@ -273,7 +273,7 @@ class PicletGeneratorService:
             print(f"Generating caption for image...")
             result = await client.predict(
                 "/stream_chat",
-                image_data,  # input_image
+                image_pil,  # input_image (PIL Image)
                 "Descriptive",  # caption_type
                 "medium-length",  # caption_length
                 [],  # extra_options
@@ -816,12 +816,14 @@ async def generate_piclet(image, hf_token: str) -> dict:
         # Get user profile (creates if doesn't exist)
         user_profile = PicletDiscoveryService.get_or_create_user_profile(user_info)
 
-        # Extract image path from Gradio input (already a filepath string)
+        # Load image once as PIL Image
+        from PIL import Image as PILImage
         image_path = image if isinstance(image, str) else str(image)
+        image_pil = PILImage.open(image_path)
 
         # Step 1: Generate caption
         print("Step 1/5: Generating image caption...")
-        caption = await PicletGeneratorService.generate_enhanced_caption(image_path, hf_token)
+        caption = await PicletGeneratorService.generate_enhanced_caption(image_pil, hf_token)
 
         # Step 2: Generate concept
         print("Step 2/5: Generating Piclet concept...")