On the same subject, when calculating the MD5 of an audio WAV file ffmpeg defaults to 16 bit. If the input is 24 bit it is my understanding that it needs to be explicitly reflected in the command as -c:a pcm_s24le otherwise the resulting MD5 will not be accurate. Are there comparable variations in the way ffmpeg processes still image files of different formats, bit depths, etc., that need to be accounted for in the command that is used?
On Sat, Jan 6, 2018 at 4:31 PM, Carl Eugen Hoyos <[email protected]> wrote: > 2018-01-06 20:38 GMT+01:00 *** <[email protected]>: > > > Somewhere I discovered that ffmpeg will create essence > > MD5s for supported still image formats. > > And audio and video. > (From FFmpeg's pov, an image format is a video format.) > > > I began using the following command to verify that > > the image essence did not change in photos after > > their IPTC metadata is revised. > > > Because there is no documentation of such a use case > > on the ffmpeg site nor anywhere else on the web I have > > begun to doubt that this is a valid use of ffmpeg or this > > command. > > Different possibilities exist within FFmpeg to calculate > checksums, the FFmpeg regression tests are a usage > example. > > > Is anyone else doing this? > > Many people are. > > > Even if you aren’t I would appreciate any comments or > > observations that you care to share. > > > > ffmpeg -i INPUT -map 0:v -f md5 OUTPUT.md5 > > This command line will calculate the md5sum of the first > video stream in the input file. > You may want to look at the framemd5 muxer for other > use-cases, other algorithms also exist. > > Carl Eugen > _______________________________________________ > ffmpeg-user mailing list > [email protected] > http://ffmpeg.org/mailman/listinfo/ffmpeg-user > > To unsubscribe, visit link above, or email > [email protected] with subject "unsubscribe". _______________________________________________ ffmpeg-user mailing list [email protected] http://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email [email protected] with subject "unsubscribe".
