- From: mwatson2 via GitHub <sysbot+gh@w3.org>
- Date: Tue, 23 Oct 2018 07:36:28 +0000
- To: public-css-archive@w3.org
PQ is display referred (absolute luminance) but that doesn't mean that HDR in general is. The basic problem is that on a linear [0-1.0] scale, 1.0 means something different for HDR vs SDR. In general, to composite, we need to know for a given HDR data source where the HDR peak white luminance should be relative to SDR peak white (or vice versa). Put another way, since SDR peak white is "page white" / "diffuse white" whereas HDR peak white is for specular highlights you arguably need to give the use two screen brightness controls (which Windows effectively does). If the source data is PQ the peak luminance (1.0 in linear space) is known to be 10,000 nits, but in general some other source of metadata is needed to specify either absolute or relative-to-SDR luminance. We have been able to [solve this problem](https://medium.com/netflix-techblog/enhancing-the-netflix-ui-experience-with-hdr-1e7506ad3e8) using an ICCv4 profile to specify BT.2020 colorspace and 16-bit 2.2 gamma coded pixel data and then using the ICC lumi tag to specify the absolute peak luminance of that signal. The pixel data could instead be PQ or HLG coded - using less bits per pixel - but these are not easy to include in ICC profiles because either you need to include massive LUTs or use the new calculator capability. -- GitHub Notification of comment by mwatson2 Please view or discuss this issue at https://github.com/w3c/csswg-drafts/issues/554#issuecomment-432130449 using your GitHub account
Received on Tuesday, 23 October 2018 07:36:32 UTC