W3C

- DRAFT -

High Dynamic Range (HDR) on the Web

21 Sep 2016

See also: IRC log

Attendees

Present
Chris_Lilley, W3C, markw, Leonard_Rosenthol, Adobe, Liam, Quin, Bert, Bos, (W3C), Lea_Verou, Invited_Expert_(CSS_WG), Mark_Watson, watsonm@netflix.com, Chris, Needham, mikeassenti, Mike_Assenti, mike.assenti@dolby.com, Florian, florian@rivoal.net
Regrets
Chair
ChrisL
Scribe
liam

Contents


ChrisL: there was an HDR breakout last year so the question is, what's changed.

CSS colour now has wide gamut, higher bit depth, so next obvious thing is higher luminance range

and how does that relate to the Web, do we want HDR in Web pages? What about photographs? video?

Some confusion about HDR vs wide gamut

Maybe we need a document explaining why this is important & a big mass market thing

someone: interested in video content creation, laptops, mobile, with new display capabilities

FrankOlivier: many new devices coming, so we have an HDR-ready compositor in Windows already

but how do we take all the sRGB content on the Web, over 99% is sRGB, how to blend sRGB content and HDR content

We're trying to decide, whether to do a dual mode compositor, one mode that assumes gamma 2.2 blending

lrosenth: we are OK with the gamma but we need a space

[questions about linear blending]

[[Frank gives a demo of adjacent colours being blended: the non-linear versions as done by browsers darken the blends]

<ChrisL> math in a non-additive colorspace

lrosenth: I'm on the board of the ICC, so I reaches out to the author of ICC Max (ng ICC), and we'd like to do joint work going forward

we have 3 ftf meetings a year, next in San Diego in November, would welcome participation; devdeloper day beforehand has seminars about ICC Max.

We also have remote call-in.

Then there are teleconferences as needed by various working groups

<FrankOlivier> http://www.color.org/iccmax.xalter

ChrisL: current version of ICC assumes white is white paper, model is that screen is for proofing of print

<lrosenth> ICC DevCon - http://www.color.org/DevCon/devcon2016/index.xalter

lrosenth: ICC Max was designed to included other media, othe use cases, e.g. video, not just paper

<lrosenth> (and we can arrange “complimentary” tickets for W3C members)

Bert: if you don't have one of these monitors, what's the fallback?

someone: you can either provide a separate fallback...

ChrisL: you could do media queries

Florian: there's a parallel to that for brightness, there's multiple ways to do it and pick one

<lrosenth> ICCMax reference implementation - https://github.com/InternationalColorConsortium/RefIccMAX

[discussion of PQ, gamma curves that map from 0..1; new idea was to map to 0..1 with different shape, and map the 0...1 to display device]

<ChrisL> hybrid log-gamma in BT.2100

dolby: we have an interest in HDR in Dolby Vision, using PQ; as Mark mentioned, displays are becoming more capable

and Web moving to TV monitors which are more likely to have more colour capability than laptop

[discussion about bit depth; 10 bit is current; 12-bit may be used in cinema/video]

ChrisL: trying to move away from 0..255 to 0..1 in CSS colour

comcast: HDR has gone mainstream in TV, going down to low price points. TV makers often include an HTML 5 environment but most are porting the open souces codebases, which don't have HDR

There are some TVs that have codebases developers by browser companies, e.g. androidTV and firefoxOS, so likely that HDR will be driven by when it's common in non-HDR devices

ChrisL: yes, I think this will happen, and there needs to be guidance

floria: I've been involved in porting Opera to TV, and indeed no deep work is going to happen at the port, has to be upstream

<mikeassenti> Dolby Vision white paper FYI http://www.dolby.com/us/en/technologies/dolby-vision/dolby-vision-white-paper.pdf

markw: we're looking at many years before tv platforms have anything other than sRGB [for non-4k video?]

liam: the blocker for the open source stuff right now tends to bt=e the toolkits, where the devs don't have access to calibration, to deep colour displayes etc

[q. about status of HDR in WebGL]

lrosenth: none of the common file formats on the Web support HDR today properly

<ChrisL> OpenRaster and JPEG-HDR? PNG at 16bit with extended chunk for transfer curve

<ChrisL> https://en.wikipedia.org/wiki/HDRi_(data_format)

[discussion of JPEG XR, jPEG 2K, other formats ]

<FrankOlivier> https://en.wikipedia.org/wiki/JPEG_XR

FrankOlivier: HEIF is another

<FrankOlivier> https://nokiatech.github.io/heif/examples.html

<mikeassenti> JPEG-XT https://jpeg.org/jpegxt/index.html

<FrankOlivier> https://nokiatech.github.io/heif/comparison.html

FrankOlivier: we'd like to see every p[art of OWP HDR-enabled. Images are the last holdout, comes down to a format/codec problem

We have to get to better compression

<ChrisL> need for a new, unencumberd, HDR, image format with good compression for delivery

Frank: all the main image formats are 20+ years old on the Web, time for a new one!

but it's astronomically diffcult

ChrisL: next steps....

general agreement there needs to be a backgrounds document giving guidence

e.g. doing linear-space compositing

<jihye> @liam i'm just on irc now

Also a need for a new image format to match today's hardware

sitting back isn't an option. Maybe spin up a community group

I'll proposed CGs for both things today

<cpn> I/me Chris Needham (BBC)

<MarkVickers> mark_vickers@comcast.com

<FrankOlivier> FrankOlivier franko@microsoft.com @frankolivier

<lrosenth> lrosenth@adobe.com

Liam: doc needs to target what Chris refers to as the "lightly interested developer"

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.144 (CVS log)
$Date: 2016/09/21 12:11:35 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.144  of Date: 2015/11/17 08:39:34  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/dog /document /
Succeeded: s/someone/markw/
No ScribeNick specified.  Guessing ScribeNick: liam
Inferring Scribes: liam

WARNING: No "Topic:" lines found.

Present: Chris_Lilley W3C markw Leonard_Rosenthol Adobe Liam Quin Bert Bos (W3C) Lea_Verou Invited_Expert_(CSS_WG) Mark_Watson watsonm@netflix.com Chris Needham mikeassenti Mike_Assenti mike.assenti@dolby.com Florian florian@rivoal.net
Got date from IRC log name: 21 Sep 2016
Guessing minutes URL: http://www.w3.org/2016/09/21-hdr-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.


WARNING: No "Topic: ..." lines found!  
Resulting HTML may have an empty (invalid) <ol>...</ol>.

Explanation: "Topic: ..." lines are used to indicate the start of 
new discussion topics or agenda items, such as:
<dbooth> Topic: Review of Amy's report


[End of scribe.perl diagnostic output]