PQ HDR in PNG - draft review

Hi color experts,

I have been asked to review a proposal for encoding raster graphics with 
the BT.2020 chromaticities and the PQ EOTF from BT.2100 in the PNG 
format. This is a spin-off from the Timed Text Markup Language (TTML) 
Working Group.

I'm one of the authors of PNG. It has the ability to embed ICC profiles 
(using the |iCCP| chunk) and also an older ability to specify the gamma 
and the RGB chromaticities (using the |gAMA| and |cHRM| chunks). If 
present, |iCCP |has precedence for color management-enabled 
applications, PNG also supports up to 16 bits per component. In 
principle then, this should be straightforward.

The proposal is at https://github.com/w3c/png-hdr-pq and the relevant 
ICC profile is at 
https://github.com/w3c/png-hdr-pq/blob/master/icc/ITUR_2100_PQ_FULL.icc 
(click on 'view raw' to download.

I used ICC Profile Inspector to examine the profile. I also converted it 
to XML for closer inspection.

This is an ICC 4.2 display profile, so it has a D50 whitepoint and a 
chad tag to indicate the adaptation from D65. The first thing I noticed 
is that there are no colorant tags to indicate the primaries used. 
Although the specification itself does list the chromaticities of the 
2020 primaries, they are not in the ICC profile.

The second thing I noticed was that the lumi tag indicates a peak 
luminance of 100 cd/m^2 which does not sound like HDR at all.

There are A2B0 and B2A0 tags for the transfer functions. The B curve is 
linear, the M curve has a gamma of 5 and there is a slightly sigmoidal A 
curve. I am not able to tell whether this correctly represents the 
BT.2100 EOTF and would appreciate guidance here.

Finally, having looked at the ICC profile for a while, i noticed an odd 
statement in the specification itself. Even though it says in the 
introduction  "This specification uses the existing |iCCP| chunk to 
unambiguously signal the color system of an image that uses the 
Reference PQ EOTF specified in [BT2100-1]" the name of the ICC profile 
is defined, "|ITUR_2100_PQ_FULL" |and this appears to be a normative 
requirement.

The specification goes on to say:

> Note
>
> The |gAMA| and embedded ICC profile are provided solely for 
> compatibility with processors that do not conform to this specification.
>
Eek! It looks as if a *magic string* is used to signal the image 
contents, and that string is the *name *of the ICC profile. Not only are 
the gamma and chromaticity to be ignored, but also the contents of the 
ICC profile.

This looks terrible! Please, if I have missed something in this 
analysis, point it out.

It seems clear that a vastly better way to encode BT.2100 still images 
in PNG would be to embed an ICCMax profile that correctly describes the 
EOTF and the primary chromaticities, and has a correct peak luminance 
value. I assume that the flaws noted above are due to limitations of ICC 
v.4?

Any obvious drawbacks of my proposed approach?



-- 
Chris Lilley
@svgeesus
Technical Director @ W3C
W3C Strategy Team, Core Web Design
W3C Architecture & Technology Team, Core Web & Media