I've managed to achieve this using obs-websocket and simple/dirty python script.
The obs-websocket plugin can get/set the pattern used by OBS to save recording files.
Install/configure obs-websocket plugin.
Then use obs-cli (https://github.com/leafac/obs-cli) or obswebsocket (https://pypi.org/project/obs-websocket-py/).
cli example (under Windows os, ...
Hellow. You need 4:4:4 pixel format and High444 profile in libx264 or h264_nvenc or other h264 encoder. And I guess that qp 0 is better then crf 0. For example:
-c:v h264_nvenc -qp 0 -pix_fmt yuv444p -profile:v high444p
For screenrecording gdigrab is the best way.
P.S.: I use Nvidia HWacel of my GeForce1660 Ti Max-Q. It's very fast!
Component Y/Pb/Pr is better than both S-video and composite. It supports higher resolution, and progressive scan (480p,720p,1080p). The Y is luminance and sync signaling, the Pb is blue minus, and the Pr is red minus. Since its progressive, it uses more bandwidth than the former's.
It's easy to get "component" mixed up for "composite"/&...