Hi Marton, > But how many frames do you intend to decode to make sure that you detect > closed captions reliably? Because as far as I understand we basically rely > on avformat_find_stream_info() to encounter a frame with closed captions. > So detection is based on the completeness of AVCodecParameters and > the order of packets from other streams, more precisely, sheer luck.
That's a good point. This is seeming more like a feature that is not compatible with the current models. Should the fields just be removed? Let users probe frames and make their own determinations? Alternatively, this can be localized to be an ffprobe concern in such a way that it can be given an option to enable CC detection and setting the fields values to a certain duration. ffprobe can then scan the frames for already exposed side data and set the field to 0/1. I think a similar approach can be taken with film grain. For example, ffprobe -show_streams -analyze_side_data 1 -analyzeduration 16M .... Where -analyze_side_data enables checking these frames up to the value of -analyzeduration. When enabled, it would set closed_captions and film_grain field based on its finding. (hypothetical new option) This makes all the work of exposing the fields, and setting dispositions unnecessary. Then only the patch #9 (avformat/demux: copy AVCodecContext properties in codec_close()) is beneficial here if we want to fix the dump function. What do you think? _______________________________________________ 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".
