John Stebbins: > On Fri, 2020-05-01 at 14:27 -0300, James Almer wrote: >> On 5/1/2020 2:09 PM, John Stebbins wrote: >>> When extra data is updated using AV_PKT_DATA_NEW_EXTRADATA, the >>> data is >>> written plus extra space is reserved for the max possible size >>> extradata. >>> But when extradata is written during mkv_write_header, no extra >>> space is >>> reserved. So the side data update overwrites whatever follows the >>> extradata in the track header and results in an invalid file. >> >> In what scenario there's extradata during init() and then new >> extradata >> propagated in a packet side data for AAC? And should the side data >> one >> take priority over the original one? >> > > This is partially a HandBrake issue. Before ffmpeg ever had side data, > HandBrake parsed the extradata out of the aac stream in TS files > ourselves before initializing the muxer. We still do this, so > extradata is supplied by us when the muxer it initialized and again by > side data generated in aac_adtstoasc bsf. > Does this mean that HandBrake's extradata and the aac_adtstoasc bsf side-data extradata are actually one and the same?
- Andreas > So remuxing ts to mkv with ffmpeg exe does not have this problem. But > it is a problem when using the API in valid ways. > > I don't think it matters which takes priority. If stream parameters > change, there really is no way to handle in mkv since this is global > data in mkv. So perhaps just ignore side data when extradata has > already been set once? > _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".