I’ve have been pondering about it for some time now and there is a reason for it.
When I compare my animated encodes frame by frame, I see that some frames come out with blurry mess but others come out better than HEVC at equivalent quality. So am I supposed to compare not still frames but frames in motion to quantify the fidelity in AV1? Or is that an area where the encoder still need improvements?
I’ve also experimented with temporal filtering on and off, sometimes it looks better, other times it butchers the fidelity.
It is important to consider that a large part of encoding fidelity is the encoder used, not the encoder format. Are you referring to svt here, or some gpu hardware encoder?
I’m refering to svt.