Implement support for "BBC OETF for Enhanced and High Dynamic Range".

[KelSolaar]

References

• Borer, T. (2014). BBC Research & Development White Paper WHP283, (July), 1–24.
• http:https://downloads.bbc.co.uk/rd/pubs/whp/whp-pdf-files/WHP283.pdf

[jaguarondi]

The BBC worked with NHK to develop the HLG (Hybrid Log Gamma) which is now standardised as "ARIB STD-B67" as described in #264

This paper is one of the steps of this research and the values were different than what they settled with in the standard. A follow-up to this from September 2015:

• http:https://downloads.bbc.co.uk/rd/pubs/whp/whp-pdf-files/WHP309.pdf

[sobotka]

DolbyPQ then enters into this discussion as well I suspect.

[jaguarondi]

The Dolby PQ is another transfer function for HDR which has been standardised in SMPTE ST 2084. PQ and HLG are indeed the 2 most important standards for HDR today. A notable difference though, HLG is an OETF (think BT.709) while PQ is an EOTF (think BT.1886).

[KelSolaar]

We have PQ implemented in develop although the location is bound to change as per #270.

[KevinJW]

I'll +1 for @jaguarondi comment about the difference between HLG and PQ.

HLG is a scene referred relative OETF
PQ is a display referred absolute EOTF

this is best demonstrated by the fact that in order to cross convert you need to know the intensity of the peak white in cd/m2 of the display. This is needed to determine an appropriate system gamma.

For now the best available info comes from BT.2390, though it is a bit of a mash up of information (I think this is intended), the true ITU standard is not yet published so it not publicly available, and doesn't have a number.

My opinion is that the proposed ITU function for the system gamma, is not going to work across different environments, in particular the lower intensity displays as found in cinema related applications might need a different figure than that predicted, both because the published papers suggest it (straight line fit on what is clearly a curve) and more importantly the lack of surround illumination. All the test evaluations were carried out with a constant surround (I believe the figure quoted was ~5 cd/m2)

Kevin

[KelSolaar]

I'm wondering if there is a point implementing that now that we have support for HLG as per #316, from an historical standpoint I think it is worth it but will it be used, I don't think so. I'll drop the issue priority.