hi,alex
10bit x265 encode I think that's good enough and then sd card storage device can also improve a well data rate setting compare to high data rate MLV mode
x.265 sup 4:2:0 4:2:2 4:4:4
On this basis
if ML can sup crop mode(1x1 sampling or 3x3 sampling)+x.265 encode so that it can provide bussines quality video and a good accessibility workflow
X.265 encode is open source,
HLG (Hybrid Log Gamma) / C-log Gamma might can be other user custom log gamma such as arri logC ,sony slog2/3 and so on
why doesn't have a try?
good luck,
sep30230
ARM clips can used Neon instruction set then run OpenCL decode
that's a ARM chromebook!
cpu: Exynos 5250,based 2 cores onCortex-A15
that is a highlight news on CES 2014
at least , ML first step may replace the poor founders built "base profile H.264 encoder " and the poor sampling
thanks to the poor encoder and the poor sampling ,EOS old DLSR is outdated
in fact its image/RAW performance also good enough in 2020
So I think that H.264 high profile + a good sampling should be support first that can improve H.264 workflow performance
slove the poor sampling is the most important task
then if X.265 availability enough took on old arm clips even though " 720P/1080p 24FPS "
it can be a good started on HEVC workflow development
good luck,
sep30230
You're acting like the developers didn't consider this?
@sep30230 - Canon uses hardware accelerated video encoding. CPU based would be much too slow. We can't "replace" the hardware encoding.
The hardware acceleration defines limits on which formats, bitrates and resolutions can be used. I'm not aware of any Canon EOS that supports H.265 (maybe some of the newer ones do? But I am pretty sure no Digic 4 or 5 do). Their dedicated video range does.
More likely to build a new camera LOL
Nice try
main issue is rec.709 cutting off dynamic range not h.264