2

I am new to video codec(encoding/decoding) stuff. I am trying to understand the difference between H.264(AVC) and H.265(HEVC) HDR video content. I thought HDR video 'means' it has 10 bit color depth and H.264 does not support 10 bit HDR content. I would appreciate any clarification on this matter. Thank you.

2
  • 1
    h.264 does support 10 bit content, as of 2005. In fact the High 4:4:4 Predictive Profile goes up to 14 bits. en.wikipedia.org/wiki/H.264/…
    – stib
    Feb 7, 2019 at 12:37
  • 10 bits is necessary, but not sufficient for HDR. Feb 10, 2019 at 7:13

3 Answers 3

1

You can do HDR with almost any modern codec. All it means in practice is that the boundary between the highest and lowest intensities in the image goes beyond what a non-HDR image could portray; if the colors linearly mapped to your monitor on a computer went from 0.0-1.0, then HDR could go above 1.0 (and potentially below 0.0) and be rendered in a convincing way with the right software.

H.265/HEVC (and its royalty-free competitor AV1/VP9) are more about encoding efficiency; H.265 can create files which are roughly half the size of H.264, but due to its licensing fees, the web and world-at-large has been very slow to adopt it.

The chief differences between H.264 and H.265 (the new stuff) are that H.265 supports up to 8K UHD as a resolution, which I don't believe H.264 did (and almost no one will use for the foreseeable future, if I'm reading the video market right); and has a more advanced compression scheme at the same quality level. This doesn't mean that the video will come through higher-quality, but that it will take less space to provide it with the same quality. It's a disk thing.

On the other hand, while H.264 is widely supported, H.265 is proprietary and supporting it requires paying a reasonably hefty licensing fee, which is part of why everyone isn't using it already. (It is quality, it's just proprietary.) For this reason, H.264 is still a very popular standard. While H.264 is generally 8-bit, either one can support a wide range of bit depths in theory.

H.264 can theoretically support up to 12-bit planar, which is usually plenty; unfortunately a lot of decoders (x264 included) have packed that into ten bits and sacrificed fidelity. I've seen people do 14-bit, I'm just not convinced that it mattered. This is still usually more than enough; but bit depth has little to do with HDR, as any range of bits can be HDR if it's just mapped to the appropriate intensities. If you need 14-bit, you will probably want to go with the more modern H.265, just on the basis of the reasons why you might need 14-bit... I think VP9 also supports this, but I'm not sure. The only drawback of VP9 that comes to mind is that you would have a marginally slower stream speed, but it's good enough for Netflix.

That said, a trained human eye can still only distinguish up to about 12 bits; for the everyman it's actually a lot lower. So, unless you're doing medical imaging or radio astronomy, this may be academic.

0

No, HDR has nothing to do with 10 bit, 8 bit can also be HDR, in both files and in displays, Nvidia/Windows supports that.

HDR is PQ or HLG transfer and those transfers are HDR.

H.264 supports even 14 bit. That is not a problem.

You should see that PQ transfer the file tagged with in Mediainfo. That means HDR is in use. Same about AVIF images.

0

Yes, it has to do with the codec. If you want to compress HDR10/10+ content, in H264 or HEVC, there are SPS & PPS markers that indicate what the HDR10 settings are within the elemental bitstream.

The metadata merely indicates standard being used : HDR10, HDR10+. While the video is simply 10-bit YUV 4:2:0 or YUV 4:2:2 typically.

After being received over the network, a decoder can extract this HDR metadata and pass it to the display for actual rendering (with full HDR effects).

The HDR10 is a specification that acts like an umbrella for the metadata coming from various properties like HLG/PQ, Brightness Nit Values, Color Primaries etc. etc. However, HDR10 is static metadata, while HDR10+ or Dolby Vision are dynamic in their metadata meaning it can change from one frame to another.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.