Re: ICC profile with headroom above SDR was: HTML Canvas - Transforms for HDR and WCG

Here is a simple use case.
Camera or other source delivers images with this kind of ICC profile.
Uploaded to Adobe Stock, Lightroom Web, Dropbox or Google or whatever.
User browses their image library, determines which images are OK vs crap.
Reasonable fidelity is needed, or else images will look like crap.
User edits image in image browser/editor.
Do image browsers/editors use canvas?

Right now, ungraded HDR images come across as too dark in Lightroom Web. 
Seems color management info is missing and it's read as SDR.

Lars

-----Original Message-----
From: Pierre Lemieux <pal@sandflow.com>
Date: Sunday, May 9, 2021 at 7:21 PM
To: Lars Borg <borg@adobe.com>
Cc: Christopher Cameron <ccameron@google.com>, Chris Lilley <chris@w3.org>, "public-colorweb@w3.org" <public-colorweb@w3.org>
Subject: Re: ICC profile with headroom above SDR was: HTML Canvas - Transforms for HDR and WCG

Hi Lars,

I am trying to understand if/how the existence of these images/ICC
profile combinations affect the Canvas API. It does not look like it
does, but maybe?

I have added the topic to the agenda of our call on 2021-05-10.

Best,

-- Pierre


On Sat, May 8, 2021 at 6:49 PM Lars Borg <borg@adobe.com> wrote:
>
> If you have specific questions, that would help.
>
> -----Original Message-----
> From: Lars Borg <borg@adobe.com>
> Date: Saturday, May 8, 2021 at 3:49 PM
> To: Pierre Lemieux <pal@sandflow.com>
> Cc: Christopher Cameron <ccameron@google.com>, Chris Lilley <chris@w3.org>, "public-colorweb@w3.org" <public-colorweb@w3.org>
> Subject: Re: ICC profile with headroom above SDR was: HTML Canvas - Transforms for HDR and WCG
>
> You're asking for use cases.
> That will take more time to determine.
>
> -----Original Message-----
> From: Pierre Lemieux <pal@sandflow.com>
> Date: Friday, May 7, 2021 at 12:40 PM
> To: Lars Borg <borg@adobe.com>
> Cc: Christopher Cameron <ccameron@google.com>, Chris Lilley <chris@w3.org>, "public-colorweb@w3.org" <public-colorweb@w3.org>
> Subject: ICC profile with headroom above SDR was: HTML Canvas - Transforms for HDR and WCG
>
> Hi Lars,
>
> It is still not clear to me how these profiles are used and how the
> images are created.
>
> Couple of questions come to mind:
>
> - are the images created on an SDR or HDR display?
>
> - is the ICC profile applied while the image is created, i.e. is the
> author manipulating the results of the image after processing through
> the ICC profile?
>
> - in the case of a PNG file, is the image expected to be acceptable on
> an sRGB display without the ICC profile applied?
>
> Best,
>
> -- Pierre
>
> On Fri, May 7, 2021 at 12:15 AM Lars Borg <borg@adobe.com> wrote:
> >
> > In this context any ICC profile with headroom above SDR white is applicable.
> > Adobe currently ships such profiles, all having a significant headroom. Also attached.
> > This covers both cine camera profiles and HDR profiles. All are V4 profiles.
> > AFAIK, there are more such profiles in the market place.
> > (Adobe started shipping such profiles in 2006)
> >
> > ARRI LogC3 Wide Color Gamut - El 160
> > ARRI LogC3 Wide Color Gamut - El 1600
> > ARRI LogC3 Wide Color Gamut - El 800
> > Canon Cinema CLog2
> > Canon Cinema CLog3
> > e-sRGB
> > P3 D65 PQ W203
> > P3 D65 PQ W300
> > P3 PQ Display Full
> > Panasonic V-Gamut V-Log
> > Rec.2100 HLG Scene W63
> > Rec.2100 HLG Scene W75
> > Rec.2100 HLG Scene W81
> > Rec.2100 HLG W100
> > Rec.2100 HLG W203
> > Rec.2100 HLG W300
> > Rec.2100 PQ W100
> > Rec.2100 PQ W203
> > Rec.2100 PQ W300
> > Rec.lTU-R BT.2100 HLG Scene Full
> > Rec.lTU-R BT.2100 HLG Scene Narrow
> > Rec.lTU-R BT.2100 PQ Display Full
> > Sony S-Gamut3.Cine/S-Log3
> > Sony S-Gamut3/S-Log3
> > Universal Camera Film Printing Density
> >
> >
> > A user can embed any of the above profiles in these file formats with these extensions (using Photoshop).
> > Other tools can support other formats.
> >
> > psd     Photoshop
> > psb     Large Document Format
> > eps     Photoshop EPS
> > jpg     JPEG
> > jpf     JPEG 2000
> > jps     JPEG Stereo
> > mpo     Multi-Picture Format
> > pdf     Photoshop PDF
> > png     PNG
> > tif     TIFF
> >
> > Not tested if these work with SVG.
> >
> > ICC profiles provide a color transform from RGB to XYZ D50.
> > ICC profiles do not provide colorimetry info in the style of an ISO 22028 color image encoding specification.
> > These profiles do not do tone mapping, but preserve the headroom.
> > The relative colorimetric intent, present in all of these profiles, together with the chad tag, provides colorimetric values for each RGB value.
> > A few of the profiles indicate scene-referred colorimetry.
> > A few of the profiles indicate nits for graphics white.
> > A few of the profiles contain the draft CICP tag.
> >
> > The ICC profile is embedded in the wrapper, not in the codec data (AFAIK).
> > Besides the ICC profile, there is no colorimetry info in the files.
> >
> > Lars
> >
> > -----Original Message-----
> > From: Pierre Lemieux <pal@sandflow.com>
> > Date: Monday, May 3, 2021 at 5:07 AM
> > To: Lars Borg <borg@adobe.com>
> > Cc: Christopher Cameron <ccameron@google.com>, Chris Lilley <chris@w3.org>, "public-colorweb@w3.org" <public-colorweb@w3.org>
> > Subject: Re: HTML Canvas - Transforms for HDR and WCG
> >
> > Hi Lars,
> >
> > I think the group will need details before the group can take any
> > meaningful action:
> >
> > - file format(s) used
> > - codec(s) used
> > - colorimetry metadata within the codec (if any)
> > - colorimetry metadata within the file (if any)
> > - colorimetry metadata within the ICC profile
> > - etc.
> >
> > Best,
> >
> > -- Pierre
> >
> > On Mon, May 3, 2021 at 12:13 AM Lars Borg <borg@adobe.com> wrote:
> > >
> > > The images with hlg and pq encoding and icc profiles already exist. Started several years ago afaik. If you need to know I can find out when we shipped our first such profile.
> > > Doing an hdr canvas with these images would be new.
> > >
> > > Thanks,
> > >
> > > Lars Borg
> > > Adobe
> > > Hawaii
> > >
> > > ________________________________
> > > From: Pierre-Anthony Lemieux <pal@sandflow.com>
> > > Sent: Sunday, May 2, 2021 2:29:04 PM
> > > To: Lars Borg <borg@adobe.com>
> > > Cc: Christopher Cameron <ccameron@google.com>; Chris Lilley <chris@w3.org>; public-colorweb@w3.org <public-colorweb@w3.org>
> > > Subject: Re: HTML Canvas - Transforms for HDR and WCG
> > >
> > > > Legacy.
> > >
> > > I do not see legacy playing a role here since HDR is new to the web.
> > >
> > > On Sun, May 2, 2021 at 5:03 PM Lars Borg <borg@adobe.com> wrote:
> > > >
> > > > Legacy.
> > > >
> > > > Lars
> > > >
> > > > -----Original Message-----
> > > > From: Pierre Lemieux <pal@sandflow.com>
> > > > Date: Saturday, May 1, 2021 at 2:26 PM
> > > > To: Christopher Cameron <ccameron@google.com>
> > > > Cc: Chris Lilley <chris@w3.org>, "public-colorweb@w3.org" <public-colorweb@w3.org>
> > > > Subject: Re: HTML Canvas - Transforms for HDR and WCG
> > > > Resent-From: <public-colorweb@w3.org>
> > > > Resent-Date: Saturday, May 1, 2021 at 2:26 PM
> > > >
> > > > > To the original question of "what to do about ICC profiles specifying HLG anonymously?".
> > > >
> > > > What is the use case for not specifying the pixel color scheme?
> > > >
> > > > Best,
> > > >
> > > > -- Pierre
> > > >
> > > > On Sat, May 1, 2021 at 5:19 PM Christopher Cameron <ccameron@google.com> wrote:
> > > > >
> > > > > On Thu, Apr 29, 2021 at 10:34 AM Chris Lilley <chris@w3.org> wrote:
> > > > >>
> > > > >>
> > > > >> On 2021-04-28 15:23, Christopher Cameron wrote:
> > > > >> > In all browsers today, for "Colorbars in HLG 203 display.png", any
> > > > >> > signal value above 0.75 is clamped to SDR-white.
> > > > >>
> > > > >> The choice of 0.75 follows from the HLG specification - a code value of
> > > > >> 0.75 is media white, while a code value of 1.0 is peak white and is 12x
> > > > >> higher luminance.
> > > > >>
> > > > >> The choice of clamping all HDR highlights, with no tone mapping is an
> > > > >> easy but bad one. However, dislaying HDR content on SDR is in general
> > > > >> hard; HLG is easier here than PQ (by design).
> > > > >
> > > > >
> > > > > Suppose we have a video that identifies itself as HLG (not anonymously), and is displayed using a <video> tag on an SDR device.  Should it clamp values above 0.75? My answer is "no".
> > > > >
> > > > > We have established that it is inevitable that pixel values over 0.75 in an image with the ICC profile "Colorbars in HLG 203 display" will be clamped.
> > > > >
> > > > > To the original question of "what to do about ICC profiles specifying HLG anonymously?". My answer would be "treat them as extended SDR, not as HLG, and there is no way for them to truly be the same as HLG".
> > > > >
> > > > >
> > > >
> >