[FFmpeg-user] Determine timecode of errors

Er Win 86Legacy at outlook.de
Tue Mar 23 00:56:54 EET 2021

Hey Mark,

nice idea, loglevel info (and up like verbose and error) gives lines like this in addition to the error log lines:

frame=80148 fps=1452 q=-0.0 size=N/A time=00:53:32.20 bitrate=N/A speed=58.2x

I think this output is shown every 0.5 seconds so depending on the coding speed it is more or less precise. Using -threads 1 the output is more fine grained (but processing takes longer, obviously). This can be used to estimate the timecode of the error (last line before and first line after the error report), however this method is not precise.
Loglevel debug does not seem to provide additional useful information about timecodes.

Yes, one can do binary search on the estimated time range using -ss and -t but there should be an easier way.

Von: ffmpeg-user <ffmpeg-user-bounces at ffmpeg.org> im Auftrag von Mark Filipak (ffmpeg) <markfilipak at bog.us>
Gesendet: Sonntag, 21. März 2021 23:59
An: ffmpeg-user at ffmpeg.org <ffmpeg-user at ffmpeg.org>
Betreff: Re: [FFmpeg-user] Determine timecode of errors

On 2021-03-21 05:04, Er Win wrote:
> Hey ffmpeg-user,
> when ffmpeg reports errors it looks like this:
> Error while decoding stream #0:0: Invalid data found when processing input
> [h264 @ 000001d8c0fe9cc0] Invalid NAL unit size (0 > 7208).
> [h264 @ 000001d8c0fe9cc0] Error splitting the input into NAL units.
> How can I find out what is the timecode for the reported error(s) in streams (audio/video)?

Have you tried 'verbose' or 'debug' reporting?
I don't know whether they would report the timecodes but it's worth a try, eh?

Mars? We can't live there! Mars is fatal: no magnetosphere to deflect cosmic rays. A tropical
sunshade orbiting Earth would work. It's the only way. We need to start construction now. Forget Mars.
ffmpeg-user mailing list
ffmpeg-user at ffmpeg.org

To unsubscribe, visit link above, or email
ffmpeg-user-request at ffmpeg.org with subject "unsubscribe".

More information about the ffmpeg-user mailing list