Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
30th October 2020, 14:24 | #4 | Link | |
ffx264/ffhevc author
Join Date: May 2007
Location: /dev/video0
Posts: 1,844
|
Quote:
I personally use in x265 a CRF of 21 with a qcomp of 0.7 and high psy-rd and psy-rdoq values. I cannot tell the difference between it and the same encoded with x264 @ CRF 18 and almost the highest possible setting (except exhaustive ME search as it's a placebo) |
|
30th October 2020, 14:28 | #5 | Link | |
SuperVirus
Join Date: Jun 2012
Location: Antarctic Japan
Posts: 1,351
|
Quote:
for x265: QP = 12 + 6 * log ₂ (qscale[i] / 0.85) source: https://www.cnblogs.com/lakeone/p/5436481.html But NOTICE: |
|
30th October 2020, 15:54 | #6 | Link |
Lost my old account :(
Join Date: Jul 2017
Posts: 326
|
At what settings? Default? Cause how can we say that crf18 in x264 equals crf20 terms of image quality in x265, when crf20 preset 'fast' will produce image quality vastly different to crf20 preset 'slower'?
In my experience slower presets uses higher bitrate at given crf-value than faster ones in x265, and the opposite is true for x264. So TS question can only be answered (at best) at specific encoder settings. |
30th October 2020, 16:24 | #7 | Link | |
ffx264/ffhevc author
Join Date: May 2007
Location: /dev/video0
Posts: 1,844
|
Quote:
|
|
30th October 2020, 21:06 | #8 | Link |
Moderator
Join Date: Jan 2006
Location: Portland, OR
Posts: 4,771
|
Also, HEVC has a lot of tools to encode more efficiently or to suppress encoding artifacts that H.264 doesn't have. So, depending on the content, x265 can look better than x264 at the same QP. This will be somewhat content dependent. I'd think x265 anime using --tskip would look at lot better at the same QP than x264. But for noisy film grain content, the gap might be smaller or non-existent. In x264 a 10-bit QP looks better than the same QP at 8-bit, but HEVC doesn't have nearly the same gap.
So, at a high level, starting with the same CRF is a decent ballpark, but I don't think we're going to have anything substantially more accurate without specifying parameters and content. I'd probably start an x265 encode tuning with the same CRF as x264, and go from there. Or maybe start it at 2 higher if I think the CRF might have been overkill in some sequences in x264. |
|
|