12 Feb
2025
12 Feb
'25
1:35 p.m.
On Wed Feb 12 00:55:33 2025 +0000, Elizabeth Figura wrote:
but i guess then there would be the problem that do we know the parser
would always break the buffer up reasonably? (e.g. one chunk per frame). what if the parser also spit out enormous buffers. In practice parsers should output individual framed buffers; that's part of the point of the parser.
so i have no idea how h264 works so i had to do some research. but looks like h264parse splits thing into what are called NALUs, not frames. in this case, NALUs are actually sub-frame (at most a frame), so we are good for this particular encoder.
but in general it feels dangerous to assume parser will always break buffer into frames or sub-frames. but maybe that's just me.