如何获取使用FFmpeg修改帧率时被丢弃的帧编号

Absolutely, you can track which frames get dropped when converting your 30fps video to 20fps with ffmpeg! Here's a practical, step-by-step approach to capture those frame numbers:

Step 1: Extract Frame Info from the Original Video #

First, we'll log every frame's details (including its number) from your source video using the showinfo filter. Run this command:

ffmpeg -i lamp02_20210212_060804-compressed.mp4 -vf "showinfo" -f null - 2> original_frames.log

• The showinfo filter outputs metadata for each frame, including the frame number (labeled n:).
• We redirect the stderr output (where ffmpeg sends filter logs) to original_frames.log so we can parse it later.

Step 2: Extract Frame Info from the Resized Video #

Next, do the same for the 20fps video you created (xx.mp4):

ffmpeg -i xx.mp4 -vf "showinfo" -f null - 2> retained_frames.log

This log will contain only the frames that were kept during the frame rate conversion.

Step 3: Compare Logs to Find Dropped Frames #

Now we need to compare the two logs to find which frame numbers exist in the original but not in the resized video. You can use either a Python script or bash commands for this.

Option A: Python Script (Easy to Customize) #

Create a script like this (save it as find_dropped_frames.py):

# Parse original video frame numbers
original_frames = set()
with open('original_frames.log', 'r') as log_file:
    for line in log_file:
        if 'n:' in line:
            # Extract the frame number from the line (e.g., "n:123" → 123)
            frame_num = int(line.split('n:')[1].split()[0])
            original_frames.add(frame_num)

# Parse retained video frame numbers
retained_frames = set()
with open('retained_frames.log', 'r') as log_file:
    for line in log_file:
        if 'n:' in line:
            frame_num = int(line.split('n:')[1].split()[0])
            retained_frames.add(frame_num)

# Calculate dropped frames and sort them
dropped_frames = sorted(original_frames - retained_frames)

# Print and save results
print("Dropped frame numbers (0-indexed):")
for num in dropped_frames:
    print(num)

with open('dropped_frames.txt', 'w') as output_file:
    for num in dropped_frames:
        output_file.write(f"{num}\n")

Run it with:

python find_dropped_frames.py

This will print the dropped frame numbers and save them to dropped_frames.txt. Note: FFmpeg's showinfo uses 0-indexed frame numbers by default—if you prefer 1-indexed, just add +1 when extracting frame_num.

Option B: Bash Commands (No Scripting Needed) #

If you prefer using terminal commands, run these:

# Extract and sort original frame numbers
grep -o 'n:[0-9]*' original_frames.log | cut -d: -f2 | sort > original_frames.txt

# Extract and sort retained frame numbers
grep -o 'n:[0-9]*' retained_frames.log | cut -d: -f2 | sort > retained_frames.txt

# Compare to find dropped frames (comm -23 shows lines only in the first file)
comm -23 original_frames.txt retained_frames.txt > dropped_frames.txt

# View the result
cat dropped_frames.txt

Key Notes #

• FFmpeg's fps filter doesn't just drop frames at fixed intervals—it selects frames based on timestamp alignment to the target frame rate. This means the dropped frames might not follow a simple pattern, so comparing actual logged frames is the most accurate way to get the list.
• If you want to avoid reprocessing the video twice, you can combine the frame rate conversion and retained frame logging into one command:

  ffmpeg -i lamp02_20210212_060804-compressed.mp4 -vf "fps=20,showinfo" xx.mp4 2> retained_frames.log
  

  This way, you generate the resized video and its frame log in a single pass.

内容的提问来源于stack exchange，提问作者user42