On Fri, Jun 9, 2023 at 12:33 PM Artem Smorodin <[email protected]> wrote: > The behavior that adds this patch was there before, but was removed after > about 5.1.1, so I think it's safe to restore it until a better solution is > implemented.
I have to admit that I'm hesitant about introducing a patch that only works for upconversion (i.e. if you go from 59.94 to 29.97 you will lose half the caption data), as well as when it does work it produces caption blocks which are malformed and will cause issues with some downstream decoders. > Regarding a better solution: > I don't think we should change the behavior of the "-r" option, instead I > suggest moving ccfifo from libavfilter to libavutil and using it in the > ffmpeg program. In my original patch series I actually had the ccfifo stuff as part of libavutil so it could be shared across the project. However there was concern about having to maintain a public ABI for a brand new interface and thus we decided to keep it private for now (and even with private APIs there is a mechanism to share them across multiple libav* libraries). That said, I hadn't considered the notion of fftools needing access to it, and I think the general trend has been to move away from fftools using private APIs. I suspect we could get "-r" to transparently add the FPS filter to the end of the filtergraph, just like we do with the vf_scale when users specify "-s" on the command line. Although this does raise the issues that Anton raised regarding dropped/duplicated frames. My inclination would be to tell people "Don't use '-r' if you care about captions being properly preserved. Use '-vf fps' instead." But I recognize that there are plenty of people who won't know the difference. Devin -- Devin Heitmueller, Senior Software Engineer LTN Global Communications o: +1 (301) 363-1001 w: https://ltnglobal.com e: [email protected] _______________________________________________ ffmpeg-devel mailing list [email protected] https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email [email protected] with subject "unsubscribe".
