simple_simon
24th December 2022, 02:02
My general guideline for encoding is to allocate 0.006MB per pixel using a 2-pass encode. I also sometimes use CRF=20 on some things which is pretty much indistinguishable to my eye from the original. The problem is if the source has a complex grain pattern then CRF=20 can create a file bigger than I really want to do. But with less complex sources CRF=20 can bring in a smaller file than my 2-pass encode and still look indistinguishable. Is there a way to do a 2-pass encode that is a regular fast 1st pass but also examines the file to decide if CRF=20 will give a smaller file size than the given 2-pass bitrate, if so then encode CRF=20, if not then encode 2nd pass using the given bitrate? It would add an additional pass to CRF encodes but it would still be faster than encoding the file both ways to determine the best way to do it.