You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I briefly examined the actual H.264 bytestream via aizvorski/h264bitstream and found some suspect things. I believe the video samples I store and the .mp4s I serve include sps/pps information, for example, even though this is redundant with the video_sample_entry.
The text was updated successfully, but these errors were encountered:
I now think this issue is fairly easy but also not very urgent.
I once thought these unnecessary NALs were preventing .m4s files from playing back properly (which would cause problems for a scrub bar UI or live view). But I found other causes for those problems, and live view is working fine now.
moonfire_nvr::h264::transform_sample_data sees each NAL and can skip ones it doesn't like. It could skip just the SPS/PPS or any "non-VCL" NALs. (See ISO/IEC 14496-10 table 7-1.) That'd save a little bit of disk space/bandwidth, but it's pretty minor. I'm not sure if any of the non-VCL NALs have info we might want to make use of in the future (maybe some cameras include in SEI some buffering info that might help us determine how far we are behind live?), but video should play fine without them, and I'm pretty sure nothing is using them today.
scottlamb
changed the title
extraneous H.264 NALs
sample data includes unnecessary non-VCL NALs
Apr 11, 2021
scottlamb
changed the title
sample data includes unnecessary non-VCL NALs
strip unnecessary non-VCL NALs from sample data
Apr 11, 2021
I briefly examined the actual H.264 bytestream via aizvorski/h264bitstream and found some suspect things. I believe the video samples I store and the
.mp4
s I serve include sps/pps information, for example, even though this is redundant with thevideo_sample_entry
.The text was updated successfully, but these errors were encountered: