[argyllcms] Re: Display P3
- From: Chris Lilley <chris@xxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Tue, 25 Jun 2019 12:01:43 -0400
On 2019-06-24 12:03, Pascal de Bruijn wrote:
On Mon, Jun 24, 2019 at 4:13 PM Chris Lilley <chris@xxxxxx
<mailto:chris@xxxxxx>> wrote:
On 2019-06-23 18:33, Graeme Gill wrote:
> Pascal de Bruijn wrote:
>
> Hi,
>
>> It might be nice to include an ICCv2 replica of Display P3 with
>> Argyll, for purposes of printer profile gamut mapping and such...
>
> should already be in ref/ as SMPTE431_P3.icm
That would be DCI P3, I imagine.
Does anybody know why Apple went with Display P3?
Because it closely matches the actual wide-gamut displays they are
shipping in their laptops and phones. And because they can use it
without restriction (see below).
It seems a bit redundant next to AdobeRGB, as it's very similar, but
not quite identical in any way (except wtpt I guess)...
Most RGB profiles use D65 whitepoint. ProPhotoRGB and DCI P3 are common
exceptions.
As far as I can see, on a technical level, Image P3 and Adobe® RGB
(1998) have identical red primary, blue primary and white
chromaticities (these are all the same as sRGB, too).
The green primary is different. They differ (slightly) in transfer
curve, white luminance (although the 80 cd/m^2 specified by both sRGB
and Image P3 is a fiction which is not observed in practice), and the
surround and viewing environment for Adobe® RGB (1998) seems more
tightly specified.
There is a major non-technical difference: trademark, registered
trademark, license agreement.
If you go to the ICC website and look at their registry of three
component encodings:
http://www.color.org/chardata/rgb/rgb_registry.xalter
you can follow the link to DCI P3 (the ancestor of Image P3)
http://www.color.org/chardata/rgb/DCIP3.xalter
and it tells you right there what the primary chromaticities are. It
also tells you that these were copied from a standard: SMPTE
EG0432-1:2010 and you can download two ICC profiles (one for D65, one
for the DCI greenish whitepoint).
now follow the link to the Adobe colorspace.
http://www.color.org/chardata/rgb/adobergb.xalter
There is no chromaticity info there (although it does describe other
things, such as the viewing conditions) and no ICC profile link.
Instead, under RGB primaries, there is a link to the Adobe website:
https://www.adobe.com/digitalimag/adobergb.html
which tells you that all Adobe products have the Adobe RGB (1998) ICC
color profile. There is a link to it,
https://www.adobe.com/support/downloads/iccprofiles/iccprofiles_win.html
(arbitrarily choosing the windows link) and if you assert that you are
not bundling this but are an end user, you get to a second link
https://www.adobe.com/support/downloads/iccprofiles/icc_eula_win_end.html
which contains trademark information and a license agreement which you
must accept (!) before it takes you to the actual ICC profile download
site, which embeds an FTP ID and won't let you download without one
https://supportdownloads.adobe.com/detail.jsp which won't work; for me
it was
https://supportdownloads.adobe.com/detail.jsp?ftpID=4075
which finally takes me to
https://supportdownloads.adobe.com/thankyou.jsp?ftpID=4075&fileID=3790
with the actual "download now" button, which links to
http://download.adobe.com/pub/adobe/iccprofiles/win/AdobeICCProfilesCS4Win_end-user.zip
which, when I tried it just now, tells me
*Not Found**
**
**The requested URL
/pub/adobe/iccprofiles/win/AdobeICCProfilesCS4Win_end-user.zip was not
found on this server.*
I have however downloaded it in the past. And looked in the profile to
see the primary chromaticities and also the chad tag, with which I can
adapt the primary chromaticities back from D50 to D65.
Oh,
https://www.adobe.com/digitalimag/adobergb.html also links to a PDF
https://www.adobe.com/digitalimag/pdfs/AdobeRGB1998.pdf
which is the actual definition, including in section 4.3.1.1 the primary
chromaticities, which must not be reproduced. However, all that data is
merely useful information for compatible encodings, or discussion,
because section 5 says
> For compliance with this specification the use of the AdobeRGB(1998)
color image encoding shall be specified by using the AdobeRGB(1998) ICC
profile.
So, they state, the only way to use this color space is to use the
Adobe-supplied ICC profile, which requires accepting a license agreement.
This is why CSS Color 4 defines a colorspace called "a98rgb" and then
notes, in passing,
> a98rgb is compatible with Adobe® RGB (1998).
and (I imagine) is why Apple picked a similar, but not identical
encoding, based on an open broadcast standard but altered to fit a
normal office/home viewing environment rather than a cinema environment,
which can be used without restriction.
Incidentally, I think that means that EXIF (which allows a choice of
either sRGB or Adobe RGB) does not conform to the Adobe document either.
Sorry for the long digression, I thought "why not use Adobe RGB"
deserved a serious answer.
What's interesting there's no bkpt tag...
Of course; this is a V4 profile. V2 ones shouldn't have it either. From
http://www.color.org/v2profiles.xalter
To avoid inconsistent results where v2 profiles are used in workflows
where both v2 and v4 profiles may be encountered, it is also
recommended that optional tags in the v2 specification that have
subsequently become obsolete are not included in future profiles made
to the v2 specification. These tags are:
* crdInfoTag.
* deviceSettingsTag
* *mediaBlackPointTag*
* namedColorTag
* ps2CRD0Tag
* ps2CRD1Tag
* ps2CRD2Tag
* ps2CRD3Tag
* ps2CSATag
* ps2RenderingIntentTag
* screeningDescTag
* screeningTag
* ucrbgTag
--
Chris Lilley
@svgeesus
Technical Director @ W3C
W3C Strategy Team, Core Web Design
W3C Architecture & Technology Team, Core Web & Media
Other related posts:
