-
-
Notifications
You must be signed in to change notification settings - Fork 308
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
x265: don't subsample chroma by default #1424
Comments
It would be useful to get a survey of what hardware and software out there currently supports 4:4:4 for hevc (and av1). Historically not much but open source supports more than the bare minimum profiles (which fortunately more often means 10-bit instead of 8-bit now, at least), and defaults need to ensure wide interoperability. |
Yes, that is one problem that may or may not exist with this. From experience, it seems realistic to be pessimistic about support for 4:4:4 even among the most popular devices. I'd like to be proven wrong, so here's my contribution to that: unexpectedly, my iOS 16 iPhone supports it just fine! Just to make it extra convenient, here's the command to use to make an image one can test (for anyone who'd like to help with compatibility testing): |
Oh, forgot to mention, psy-rd is default enabled in x265 but it's disabled by default in libheif because it sets That issue would need to be solved, otherwise 4:4:4 chroma would not get halved quality by default. |
Do you have metrics for each option (or combination of options)? I'm interested in how those options affect both resulting quality (objective and subjective) and file size. That could help to inform a change. |
x265 has a section of code that automatically runs when outputting 4:4:4 that obtains the bandwidth reduction of 4:2:0 chroma subsampling but without the downsides of chroma subsampling even with sharp YUV (which still adds a lot of glowy artifacts in addition to crushing fine details).
Letting x265 encode to 4:4:4 by default wouldn't meaningfully increase file size (unless the chroma is very complex, the size difference is negligible compared to 4:2:0), but it would mean it can handle cases that chroma subsampling definitionally cannot.
The text was updated successfully, but these errors were encountered: