On 8/13/2018 7:58 PM, James Almer wrote: > On 8/13/2018 7:47 PM, Carl Eugen Hoyos wrote: >> 2018-08-14 0:36 GMT+02:00, Thomas Daede <bztdli...@gmail.com>: >>> On 08/13/2018 03:30 PM, Carl Eugen Hoyos wrote: >>>> Why isn't av1 added like the existing codecs? >>> >>> That would generate invalid AV1 files if -strict -1 or -2 specified >> >> That is the idea of -1 and -2. >> >>> (and currently AV1 mov writing is behind -2). >> >> Is the specification still unfinished? > > Yes. > >> Then what is the exact purpose of this patch? > > The issue at hand is that when remuxing from ivf, the propagated codec > tag AV01 is written as is instead of the correct av01 (as listed in > ff_codec_movvideo_tags and codec_mp4_tags) because a case insensitive > check takes place in either ffmpeg.c or libavformat generic code (i > don't recall exactly where), which of course assumes av01 == AV01. > The latter is only correct for ivf, not mp4. > > If this patch adds the AV1 codec id inside the "s->strict_std_compliance >> = FF_COMPLIANCE_NORMAL" check as you suggest, and a muxing process is > done with -strict experimental, then the wrong codec_tag will be written. > > An alternative would be to change the aforementioned case insensitive > check into a case sensitive one, but that could break some other > scenarios as i presume it was made that way for a reason.
If you're curious why this doesn't happen with VP9, another format supported by both ivf and mp4, it's because ivf is VP90 and mp4 is vp09. _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
