video conversion bit-ness

[justinrh 259]

What is good practice when converting video with respect to bit-depth?  Is it best to match or does increasing it do any good?  I usually go from MPEG to AVC or HEVC.

Note:  I just did a test with HandBrake where my source was 8-bit, then converted to HEVC with 10-bit, 12-bit, and 8-bit.  The 10-bit was the largest file size and the others about 100MB smaller.

[Luke 42077]

HI, there is no reason to increase it when converting. I would suggest matching to preserve data.

[Suliamu 36]

x265/HEVC behaves differently than x264/AVC and there is a lot of subjectivity involved in assessing the results, but here in this article
the author argues strongly for converting 8bit to 10bit HEVC, at least for Anime quote:

Quote

To keep things short and not get into technical details: use the 10-bit encoder (Main 10 profile).

10-bit produces slightly smaller files while preventing banding from 8-bit compressions (this isn’t a joke, quantization and linear algebra is a mysterious thing).

Why not just use 12-bit then you say? Well, to put simply: moving form 8 to 10 bit increases color gradient available from 256 to 1024 to eliminate banding, but going to 4048 in 12-bit really isn’t noticeable. In fact, it is much less supported, and from my tests back in 2018 the 12-bit encoder is actually worse than the 10-bit encoder at high crf likely due to less resources put into developing it.

https://kokomins.wordpress.com/2019/10/10/anime-encoding-guide-for-x265-and-why-to-never-use-flac/#which-x265-encoder-8-bit-10-bit-or-12-bit

[justinrh 259]

Thanks for that info.  That makes sense with why HB defaults to 10-bit for HEVC.  Didn't expect 10-bit to create a smaller file ...