How to Stop FFmpeg from Doing Anything with the Supplied Video, Other than Dumping Metadata

To extract metadata without altering the video file, use FFmpeg with the -f lavfi input format and amovie=video.mp4 to read the stream as a filter source. By piping this through astats or showinfo filters and outputting to null, you ensure zero encoding occurs, preventing any modification to the original media while allowing detailed inspection of headers and stream properties.

Understanding Metadata Extraction vs. Transcoding

Many users assume that simply asking FFmpeg to copy streams will preserve files, but standard transcoding workflows often re-encode audio or video even if the codec is set to copy. This can lead to subtle changes in timestamps, packet ordering, or container structure. When the goal is strictly diagnostic—such as verifying frame rates, checking for corrupt headers, or dumping all available tags—the safest approach is to avoid the muxer entirely. FFmpeg’s architecture separates demuxing (reading) from muxing (writing). To stop it from doing anything to the supplied video other than reading its information, we must bypass the normal encoder/decoder pipeline. Instead, we treat the input as a data source for analysis filters. This method ensures that no new packets are generated for the output container, effectively making the operation read-only regarding the final file structure. For related tasks involving visual inspection, you might also be interested in FFMpeg Image Creation from Timestamps, which shows how to capture specific frames when you need visual confirmation alongside metadata.

The Correct Command Structure

The most reliable way to dump metadata without touching the video content is using the lavfi (Libavfilter Input) protocol. This allows you to create virtual sources from existing files. The core concept is to load the video file into a filter graph, analyze it, and then discard the result. Here is the correct command syntax:

ffmpeg -f lavfi -i "amovie=video.mp4,showinfo" -an -vn -sn -y null

Let us break down why each part is necessary. The -f lavfi flag tells FFmpeg that the input is not a standard media file but a filter graph description. The -i argument defines the source. Inside the quotes, amovie=video.mp4 loads the actual video file into the filter system. The ,showinfo part attaches a filter that prints detailed information about every frame and packet to the standard error output. The flags -an, -vn, and -sn disable audio, video, and subtitle outputs respectively. Finally, null is a built-in FFmpeg muxer that discards all incoming data. Because no data is written to disk, the original file remains completely untouched. The -y flag overwrites any existing output files if they exist, though in this case, since we are writing to null, it simply prevents interactive prompts. If you only care about the header information (tags, duration, codec names) rather than per-frame details, you can simplify this by using the default behavior with -hide_banner and redirecting output, but the lavfi method is more robust for deep inspection.

Alternative: Using ffprobe

While FFmpeg works, the dedicated tool ffprobe is often better suited for pure metadata extraction. It is designed specifically for this purpose and does not require complex filter graphs.

ffprobe -v error -show_format -show_streams video.mp4

This command outputs JSON-like structured data containing all stream details and format tags. It is faster and less resource-intensive than running the full FFmpeg binary. However, if you need to verify that the file is playable or check for specific filter compatibility, the FFmpeg lavfi method provides a more comprehensive health check. For users who frequently deal with large batches of media, understanding how to handle timing issues is crucial. You may find Muxing AC3 with x264, timing & seeking challenges useful when you eventually move from inspection to actual processing.

Common Pitfalls and Troubleshooting

A common mistake is omitting the -an, -vn, or -sn flags. If you leave them out, FFmpeg will attempt to decode and re-encode the streams before sending them to the null muxer. This defeats the purpose of avoiding manipulation, as decoding introduces potential quality loss or timestamp adjustments depending on the decoder implementation. Another issue arises with the path to the video file. If your filename contains spaces, you must enclose the entire input string in quotes. For example:

ffmpeg -f lavfi -i "amovie='My Video File.mp4',showinfo" -an -vn -sn -y null

Failure to quote correctly will result in an "Invalid data found" error because FFmpeg interprets the space as a delimiter between arguments. Additionally, some older versions of FFmpeg may behave differently with the amovie filter. If you encounter errors, ensure you are using a recent version of FFmpeg. The lavfi input method has been stable for many years, but filter syntax can change. Always check your version with ffmpeg -version.

Performance Considerations

Running showinfo on high-resolution videos can generate massive amounts of text output. Each frame produces several lines of data. For a 4K video at 60fps, this can overwhelm terminal buffers or slow down disk I/O if you redirect the output to a file. In such cases, consider using volumedetect for audio-only analysis or limiting the duration with -t to test a short segment first. If you are extracting thumbnails for preview purposes, note that metadata extraction does not generate images. For that, refer to FFmpeg Series Thumbnails Created Less Than Expected to understand how to properly configure frame extraction without corrupting the source.

Verifying Success

To confirm that no changes were made to the original file, compare the file hash before and after running the command. Use md5sum or sha256sum on Linux/macOS, or a checksum utility on Windows. If the hash remains identical, the process was truly non-destructive. You can also check the file modification timestamp. While the content is unchanged, some systems update the access time. This is normal and does not indicate corruption. The key is that the binary content of the file remains exactly as it was before the command started.

Frequently Asked Questions

Why does my metadata extraction still take a long time?

Even though no video is encoded, FFmpeg must still demux and parse every frame to feed the showinfo filter. For very long or high-bitrate files, this parsing process is CPU-intensive. Using ffprobe instead is significantly faster as it skips frame-level analysis unless explicitly requested.

Can I extract metadata for multiple files at once?

Yes, you can use a loop in bash or PowerShell to iterate through files. For example, in bash, you can use for f in *.mp4; do ffmpeg -f lavfi -i "amovie=$f,showinfo" -an -vn -sn -y null; done. Be careful with variable expansion and quoting to avoid errors with special characters in filenames.

What if the file is corrupted and FFmpeg fails to read it?

If the file is severely corrupted, FFmpeg may refuse to open it entirely. In such cases, try using -recovery_enabled 1 or -err_detect ignore_err to force FFmpeg to attempt reading as much as possible. However, some corruption types make recovery impossible, and the file may need repair tools outside of FFmpeg.

Does this method work for network streams?

No, the amovie filter is designed for local files. For network streams, you would typically use -i http://example.com/stream with standard options. However, analyzing live streams requires different strategies, such as capturing a segment and then analyzing it locally.

Is there a way to get just the tags without frame info?

Yes, you can use the dump filter or rely on ffprobe's -show_format option. If you must use FFmpeg, you can replace showinfo with metadata=dump in the filter chain, but ffprobe remains the most efficient tool for tag-only extraction.