To concatenate video files while respecting specific outpoint timestamps and preserving original codec parameters like bitrate and time_base, you must use the concat demuxer with a properly formatted text file. This approach avoids re-encoding, ensuring that the output matches the input quality exactly. The key is to define precise start and end times for each segment in the demuxer list, allowing FFmpeg to cut clips without altering their internal stream properties.
Understanding the Concat Demuxer
The concat demuxer is distinct from the concat protocol or filter because it does not require the input files to be of the same format or codec. It reads a plain text file that lists the media files and their respective durations or timestamps. When you use the
-f concat flag, FFmpeg expects this list to guide the muxing process. This method is particularly useful when you need to stitch together segments that have different resolutions or codecs, although for our purpose of preserving parameters, we focus on identical streams.
Why Streamcopy Matters
Preserving input codec parameters means avoiding re-encoding. Re-encoding often leads to generation loss, where the video quality degrades with every pass. By using
-c copy, you instruct FFmpeg to simply copy the audio and video streams from the source containers to the output container. This is fast and lossless. However, standard concatenation might ignore your desired cut points if not configured correctly. The demuxer allows you to specify exactly how much of each file to include, honoring your outpoint constraints.
Creating the Demuxer List File
The core of this technique is the input list file. This is a simple text file, often named
input.txt, that contains directives for FFmpeg. Each line represents a file to be concatenated. To honor outpoint timestamps, you can use the
file directive followed by the path to the media file. More importantly, you can append a
duration parameter to each entry. This tells FFmpeg to stop reading that specific file after the specified number of seconds.
For example, if you have two videos,
clip1.mp4 and
clip2.mp4, and you want to take only the first 5 seconds of the first clip and the entire second clip, your list would look like this:
file 'clip1.mp4'
duration 5.0
file 'clip2.mp4'
Note that the duration is specified in seconds as a floating-point number. If you omit the duration, FFmpeg will consume the entire file. This mechanism effectively acts as an outpoint marker for each segment. You can also use the
inpoint directive if you need to skip the beginning of a file, though the primary requirement here is honoring the outpoint.
Handling Timestamps and Time Base
When you preserve codec parameters, the
time_base of the output stream remains identical to the input. This is crucial for precise editing. If you were to re-encode, FFmpeg might normalize the time base, which could introduce slight timing discrepancies or frame drops. By keeping the original
time_base, you ensure that the frames align perfectly with the timestamps defined in your demuxer list. This precision is vital for professional video workflows where sync issues are unacceptable.
Executing the FFmpeg Command
Once your list file is ready, you can run the FFmpeg command. The basic structure involves specifying the format, the input list, and the output file. You must also explicitly tell FFmpeg to copy the streams.
ffmpeg -f concat -safe 0 -i input.txt -c copy output.mp4
The
-safe 0 option is often necessary if your file paths contain special characters or are absolute paths, preventing FFmpeg from rejecting them for security reasons. The
-c copy flag ensures no re-encoding occurs. If you encounter errors related to unsupported formats, verify that all input files share compatible codecs. For instance, mixing H.264 video with HEVC video in a single MP4 container via concat demuxer may fail depending on the container's capabilities, even though the demuxer itself is flexible. In such cases, transcoding one stream to match the other is required, but this violates the "preserve parameters" goal. Therefore, ensure all inputs are encoded identically.
Common Pitfalls and Solutions
One common issue is that the final output may have a slightly incorrect duration due to rounding errors in the demuxer processing. Another is that some players might struggle with discontinuous timestamps if the segments are not seamless. To mitigate this, you can force a keyframe at the cut point, but since we are using
-c copy, we cannot insert new keyframes. Instead, ensure your source files have regular keyframe intervals. If the outpoint falls between keyframes, the player might seek to the previous keyframe, causing a slight overlap or gap.
For more complex editing tasks, such as stacking videos vertically, consider exploring alternative methods that involve filters rather than simple concatenation. See
Efficient Way of Stacking Two Videos Vertically Using FFmpeg for advanced compositing techniques. Additionally, understanding the difference between
-ss,
-t, and
-to flags is essential when dealing with timestamps outside the demuxer context. Read
Understanding FFmpeg's ss, -t, -to, Streamcopy, and Keyframe Differences to deepen your knowledge of time-based operations.
Frequently Asked Questions
Can I mix different video codecs in the concat demuxer?
Yes, the concat demuxer supports mixed codecs, but the output container must support them. For example, you can concatenate H.264 and HEVC streams into a Matroska (MKV) container. However, copying streams directly may result in compatibility issues with some players, so testing is recommended.
What happens if my outpoint timestamp is not on a keyframe?
Since you are using
-c copy, FFmpeg cannot insert a new keyframe at the cut point. The resulting file will have a keyframe before the cut point. Some players may seek back to that keyframe, causing a few extra frames to appear before your intended outpoint. This is a limitation of stream copying.
How do I handle audio desynchronization during concatenation?
If the audio and video streams have different durations or timestamps, they may drift apart. Ensure that the input files have synchronized audio and video tracks. Using
-c copy preserves this synchronization. If issues persist, check the metadata of the input files for any timestamp discrepancies.
Is the concat demuxer faster than the concat filter?
Generally, yes. The concat demuxer operates at the container level, making it very fast for simple sequential joining. The concat filter operates at the frame level, allowing for more complex manipulations like crossfades or resizing, but it requires re-encoding or complex stream handling, which is slower.
Can I use relative paths in the input list file?
Yes, relative paths work fine. However, if you use absolute paths, you must use the
-safe 0 flag to prevent FFmpeg from blocking access due to security restrictions. It is best practice to keep all media files and the list file in the same directory for simplicity.