I am trying to shrink some videos from 1920x1080 to 1280x720 and to achieve a reasonable compromise between video quality and file size. I have decided to give FFMPEG's H.265 encoder a try, and now would like to know which CRF value I should use to get "visually lossless" results.
I know that this term may be misleading, but I have taken it from FFMPEG's H.264 documentation. In the section about the CRF value:
Consider 17 or 18 to be visually lossless or nearly so; it should look the same or nearly the same as the input but it isn't technically lossless.
Furthermore, in FFMPEG's H.265 documentation, in the section about the CRF value:
The default is 28, and it should visually correspond to libx264 video at CRF 23, but result in about half the file size.
So I have learned that a H.265 CRF 28 should produce roughly the same visual results as a H.264 CRF 23.
Does that mean that a H.265 CRF 22 should produce roughly the same visual results as a H.264 CRF 17?
In other words, is the following statement correct: If I encode a video with H.264 with a CRF of
N, do I get the same visual result if I encode the same video with H.265 with a CRF of
N+5, independent of
N? Or does that formula only apply for
N=23, respectively, since this is the example the FFMPEG H.265 documentation gives?