Update app.py

app.py

@@ -144,15 +144,9 @@ def navigate_pdf_page(direction: str, state: Dict[str, Any]):
     return image_preview, state, page_info_html
 
 def downsample_video(video_path):
-    """
-    Downsamples the video to evenly spaced frames.
-    Each frame is returned as a PIL image along with its timestamp.
-    """
     vidcap = cv2.VideoCapture(video_path)
     total_frames = int(vidcap.get(cv2.CAP_PROP_FRAME_COUNT))
-    fps = vidcap.get(cv2.CAP_PROP_FPS)
     frames = []
-    # Use a maximum of 10 frames to avoid excessive memory usage
     frame_indices = np.linspace(0, total_frames - 1, min(total_frames, 10), dtype=int)
     for i in frame_indices:
         vidcap.set(cv2.CAP_PROP_POS_FRAMES, i)
@@ -160,8 +154,7 @@ def downsample_video(video_path):
         if success:
             image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
             pil_image = Image.fromarray(image)
-            timestamp = round(i / fps, 2)
-            frames.append((pil_image, timestamp))
+            frames.append(pil_image)
     vidcap.release()
     return frames
 
@@ -270,6 +263,7 @@ def generate_video(model_name: str, text: str, video_path: str,
         time.sleep(0.01)
         yield buffer, buffer
 
+
 @spaces.GPU
 def generate_pdf(model_name: str, text: str, state: Dict[str, Any], 
                  max_new_tokens: int = 2048,