I'm live streaming a 1080p media (that's, I'm screencasting my desktop) and transcoding it for less-powerfull devices on wifi (like a sasmung S4 mini smartphone for example).
Here's my current stream output definition:
Code: Select all
#transcode{vfilter=canvas{width=852,height=480},vcodec=h264,venc=x264{profile=baseline,level=3,preset=ultrafast,tune=zerolatency,keyint=50},acodec=aac,ab=128,channels=2,samplerate=44100}:rtp{sdp=rtsp://:9090/desktop.sdp}
Code: Select all
#transcode{vcodec=h264,venc=x264{profile=baseline,level=3,preset=ultrafast,tune=zerolatency,crf=28,keyint=50},acodec=aac,ab=128,channels=2,samplerate=44100}:rtp{sdp=rtsp://:9090/desktop.sdp}
What do you think in general about this idea of just lowering the CRF quality?
Am I right in believing that downscaling before encoding is better (performance-wise) than simply lowering the CRF quality?
Do the conclusions in https://developers.google.com/media/vp9/bitrate-modes/ also hold true for the x264 encoder?
Thanks for the help.
Cyrille