[FFmpeg-user] 3 ffprobe (or ffmpeg) questions regarding video identification
This is a bit OT but most of the folks here are probably familiar with MPV... Regarding ffprobe & MPV's File, Video, and Audio properties (keyboard 'Shift I'), If ffprobe shows '23.98 fps, 23.98 tbr' and MPV shows 'FPS: 23.976 (specified) 23.976 (estimated)' (i.e., ffprobe & MPV agree), I understand what that means and how it was determined. However, if ffprobe shows '29.83 fps, 59.94 tbr' and MPV shows something like 'FPS: 29.833 (specified) 23.976 (estimated)', I'm at a loss on 3 points: 1, Why do ffprobe & MPV differ? and 2, How does MPV know that the video is 23.976 Hz hard telecined (versus an NTSC telecast)? and 3, Why is ffprobe displaying a 59.94 Hz time-base rate? Regarding point 2, I can imagine that MPV is doing some sort of picture analysis to determine that the bottom fields contain even-concurrent half-pictures (versus sequential half-pictures). Am I correct? Regarding point 3, does the switch from '23.98 tbr' to '59.94 tbr' indicate that ffprobe switches from reporting frame rate to reporting field rate when it encounters fields instead of frames? Most importantly, does anyone know of an ffprobe (or ffmpeg) command syntax that will reproduce what MPV is showing (i.e., '23.976' as the video origin's, or internal, frame rate)? My current ffmpeg heuristics probe command is CMD /K ffmpeg -ss 30.0 -i %1 -filter:v idet -frames:v 900 -an -f rawvideo -y NUL but it does not find what MPV is clearly able to find -- I say that because MPV is correct in it's analysis. Thanks in advance, Mark. ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
[FFmpeg-user] H.264 and RockChip MPP
Hi there, I compiled the RockChip fork of ffmpeg because I wanted to use hardware acceleration on one of their chips (RK3399). For the time being, it looks like only the hardware accelerated decoding is implemented. So I tried: ffmpeg -benchmark -loglevel 48 -vcodec h264_rkmpp -i bbb_sunflower_1080p_30fps_normal.mp4 -map 0:v:0 -f null - and got at the end: frame=19036 fps=237 q=-0.0 Lsize=N/A time=00:10:34.83 bitrate=N/A speed=7.91x video:7585kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: unknown Input file #0 (/media/linaro/0E9ACA2339F586E5/slides_rushes/bbb_sunflower_1080p_30fps_normal.mp4): Input stream #0:0 (video): 19036 packets read (237848132 bytes); 19036 frames decoded; Input stream #0:1 (audio): 21 packets read (10080 bytes); Input stream #0:2 (audio): 2 packets read (2560 bytes); Total: 19059 packets (237860772 bytes) demuxed Output file #0 (pipe:): Output stream #0:0 (video): 19036 frames encoded; 19036 packets muxed (7766688 bytes); Total: 19036 packets (7766688 bytes) muxed mpp_thread: MPP_THREAD_STOPPING status set mThread 0xab44b8f0 mpp_dec: mpp_dec_parser_thread exit mpp_dec: mpp_dec_parser_thread exit ok mpp_thread: MPP_THREAD_STOPPING status set mThread 0xab44b9f8 mpp_dec: mpp_dec_hal_thread exit ok bench: utime=10.101s 19036 frames successfully decoded, 0 decoding errors bench: maxrss=93880kB [AVIOContext @ 0xab3cd8a0] Statistics: 259311905 bytes read, 972 seeks As I had read RockChip had made a PR to ffmpeg, I tried the latest source from ffmpeg official repository, and to my surprise, I also got the RockChip hardware accelerated decoding. frame=19036 fps=234 q=-0.0 Lsize=N/A time=00:10:34.83 bitrate=N/A speed=7.79x video:7585kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: unknown Input file #0 (/media/linaro/0E9ACA2339F586E5/slides_rushes/bbb_sunflower_1080p_30fps_normal.mp4): Input stream #0:0 (video): 19036 packets read (237848132 bytes); 19036 frames decoded; Input stream #0:1 (audio): 21 packets read (10080 bytes); Input stream #0:2 (audio): 2 packets read (2560 bytes); Total: 19059 packets (237860772 bytes) demuxed Output file #0 (pipe:): Output stream #0:0 (video): 19036 frames encoded; 19036 packets muxed (7766688 bytes); Total: 19036 packets (7766688 bytes) muxed mpp_thread: MPP_THREAD_STOPPING status set mThread 0xab149dd0 mpp_dec: mpp_dec_parser_thread exit mpp_dec: mpp_dec_parser_thread exit ok mpp_thread: MPP_THREAD_STOPPING status set mThread 0xab149ed8 mpp_dec: mpp_dec_hal_thread exit ok bench: utime=9.628s stime=8.357s rtime=81.544s 19036 frames successfully decoded, 0 decoding errors bench: maxrss=75008kB [AVIOContext @ 0xab11b8b0] Statistics: 259311905 bytes read, 972 seeks I suppose there are lots of other RockChip specific features that have not made it yet into the official ffmpeg repository, but for the time being, it looks like half of what I need is already there. On the RockChip wiki, it is written that the hardware encoding should be available Q4 2019 for ffmpeg, so I still have hope (if this work, I could deploy lots of these machines in the company I work for). It seems to be already available for gstreamer. Is there anything going on on your side for the RockChip hardware encoding? Thanks. -- Bruno Verachten ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] ffmpeg cuvid not transcoding properly
No luck with -vf yadif_cuda=0:-1:0 ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-user] ffmpeg cuvid not transcoding properly
Try using yadif_cuda to deinterlace instead ___ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".