Update app.py

app.py

@@ -18,9 +18,9 @@ def extract_frames(video_path):
     cap = cv2.VideoCapture(video_path)
     frames = []
     while True:
-        ret, frame = cap.read()  # ret is True if a frame is read.
+        ret, frame = cap.read()  # ret is True if a frame is successfully read.
         if not ret:
-            break  # Stop if no more frames.
+            break  # Exit loop when no more frames.
         frames.append(frame)
     cap.release()
     return frames
@@ -32,10 +32,10 @@ def apply_style_propagation(frames, style_image_path):
     """
     Applies the style from the provided image onto each video frame.
     Logic:
-    - Resize the style image to match the video dimensions.
-    - Use the style image as the starting point.
+    - Load and resize the style image to match video dimensions.
+    - Use the style image as the first styled frame.
     - For each subsequent frame, compute optical flow between consecutive frames.
-    - Warp the previous styled frame so that the style follows the motion.
+    - Warp the previous styled frame using the flow so that the style follows the motion.
     """
     # Load and resize the style image.
     style_image = cv2.imread(style_image_path)
@@ -83,7 +83,7 @@ def save_video(frames, output_path, fps=30):
     Combines frames into a video and saves it.
     Logic:
     - Create a VideoWriter with the specified FPS and frame size.
-    - Write each frame to the video file.
+    - Write each frame sequentially to the video file.
     """
     h, w, _ = frames[0].shape
     fourcc = cv2.VideoWriter_fourcc(*'mp4v')  # Use 'mp4v' codec for MP4.
@@ -106,12 +106,24 @@ def process_video(video_file, style_image_file, fps=30):
     Returns:
       - Path to the generated styled video.
     """
-    # Get file paths from the uploaded file objects.
-    # If not a string, assume it's a dictionary with a "name" key.
+    # For the video file, we expect a file path.
     video_path = video_file if isinstance(video_file, str) else video_file["name"]
-    style_image_path = style_image_file if isinstance(style_image_file, str) else style_image_file["name"]
     
-    # Extract frames from the video.
+    # For the style image, Gradio might return a numpy array.
+    if isinstance(style_image_file, str):
+        style_image_path = style_image_file
+    elif isinstance(style_image_file, dict) and "name" in style_image_file:
+        style_image_path = style_image_file["name"]
+    elif isinstance(style_image_file, np.ndarray):
+        # If the image is a numpy array, save it to a temporary file.
+        tmp_path = os.path.join(tempfile.gettempdir(), "temp_style_image.jpeg")
+        # Gradio images are usually in RGB; OpenCV uses BGR.
+        cv2.imwrite(tmp_path, cv2.cvtColor(style_image_file, cv2.COLOR_RGB2BGR))
+        style_image_path = tmp_path
+    else:
+        return "Error: Unsupported style image file format."
+    
+    # Extract frames from the input video.
     frames = extract_frames(video_path)
     if not frames:
         return "Error: No frames extracted from the video."
@@ -145,7 +157,7 @@ iface = gr.Interface(
 )
 
 # -----------------------------
-# Launch the Gradio App.
+# Launch the Gradio App with public sharing enabled.
 # -----------------------------
 if __name__ == "__main__":
-    iface.launch()
+    iface.launch(share=True)