update

eagle_vl/serve/chat_utils.py

@@ -23,36 +23,57 @@ import tempfile
 import os
 import imageio
 
-def compress_video_to_base64(video_path: str, max_frames=128, resolution=(960, 540)) -> str:
+
+def compress_video_to_base64(
+    video_path: str,
+    max_frames: int = 128,
+    resolution: tuple = (960, 540),
+    target_crf: int = 28
+) -> str:
     cap = cv2.VideoCapture(video_path)
-    total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
+    if not cap.isOpened():
+        raise RuntimeError(f"无法打开视频：{video_path}")
+
+    total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT)) or None
+    original_fps = cap.get(cv2.CAP_PROP_FPS) or None
+
+    if not total_frames or not original_fps:
+        cap.release()
+        raise RuntimeError("无法获取视频帧数或帧率，请检查视频文件或使用 ffprobe。")
+
     step = max(1, total_frames // max_frames)
+    new_fps = max(1, round(original_fps / step))
+
+    with tempfile.NamedTemporaryFile(suffix=".mp4", delete=False) as tmp:
+        tmp_path = tmp.name
+
+    writer = imageio.get_writer(
+        tmp_path,
+        fps=new_fps,
+        codec='libx264',
+        ffmpeg_params=[
+            '-crf', str(target_crf),
+            '-pix_fmt', 'yuv420p'
+        ]
+    )
 
-    frames = []
-    count = 0
-    while cap.isOpened():
+    frame_idx = 0
+    while True:
         ret, frame = cap.read()
         if not ret:
             break
-        if count % step == 0:
-            frame_resized = cv2.resize(frame, resolution)
-            frames.append(cv2.cvtColor(frame_resized, cv2.COLOR_BGR2RGB))
-        count += 1
-    cap.release()
-
-    with tempfile.NamedTemporaryFile(suffix=".mp4", delete=False) as tmp:
-        tmp_path = tmp.name
+        if frame_idx % step == 0:
+            small = cv2.resize(frame, resolution)
+            writer.append_data(cv2.cvtColor(small, cv2.COLOR_BGR2RGB))
+        frame_idx += 1
 
-    writer = imageio.get_writer(tmp_path, fps=10, codec='libx264', quality=8)  # quality: 0(worst) - 10(best)
-    for f in frames:
-        writer.append_data(f)
+    cap.release()
     writer.close()
 
     with open(tmp_path, "rb") as f:
-        video_data = f.read()
+        data = f.read()
     os.remove(tmp_path)
-
-    return base64.b64encode(video_data).decode("utf-8")
+    return base64.b64encode(data).decode("utf-8")
 
 
 
@@ -448,11 +469,14 @@ def to_gradio_chatbot(conversation: Conversation) -> list:
                             # Fallback to link
                             media_str += f'<a href="{item}" target="_blank">{item}</a>'
                     elif isinstance(item, str) and (item.endswith((".mp4", ".mov", ".avi", ".webm"))):
-                        b64 = compress_video_to_base64(item)
-                        media_str += (
-                            f'<video controls style="max-width:300px;height:auto;" '
-                            f'src="data:video/mp4;base64,{b64}"></video>'
-                        )
+                        try:
+                            b64 = compress_video_to_base64(item)
+                            media_str += (
+                                f'<video controls style="max-width:300px;height:auto;" '
+                                f'src="data:video/mp4;base64,{b64}"></video>'
+                            )
+                        except:
+                            pass
 
                     # If PIL image
                     else: