[css-color-4] Hyphens in colorspaces

[jstblck]

Since the modified version of DCI-P3 in CSS 4 is the same as the version already in use by Apple, would it not make more sense to use the same name, ie display-p3 instead of image-p3?

From CSS 4:

It uses the same primary chromaticities as DCI-P3, but with a D65 whitepoint and the same transfer curve as sRGB.

From Apple:

This color space uses the DCI P3 primaries, a D65 white point, and the same gamma curve as the sRGB color space.

'Display P3' is also used on some forums and in places on Wikipedia with the above meaning.

Also, for consistency, should Adobe RGB (1998) be named adobe-rgb-98 or preferably just adobe-rgb, and ProPhoto RGB named prophoto-rgb?

This would make the naming scheme the common name (but with a hyphen replacing any spaces) with, if necessary, the version (such as the year for Adobe RGB) as a hyphen-separated suffix.

[svgeesus]

Thanks for your comments.

The P3 colorspace was introduced into CSS Color at the request of Apple. We originally called it DCI P3 because that was what we thought it was. As more information emerged (different whitepoint, different viewing conditions and transfer curve) the naming was discussed in the CSS WG. image-p3 is the name that emerged as the favourite. Representatives from Apple were involved in those discussions.

Adobe RGB (1998) is the official, trademarked, name of the Adobe colorspace. For licensing, trademark and copyright reasons, that phrase can't be used to describe the colorspace used in CSS. (I know). Nor can we use the phrase Adobe which would be a registered trademark infringement.

By the way there were several variants of the Adobe colorspace, so the 1998 specifier is an important part. This is why CSS has a colorspace called a98rgb.

It might indeed be more consistent to write a98-rgb and also prophoto-rgb which would also be easier to read.

[grorg]

FWIW, I prefer the name display-p3

[svgeesus]

Well, ok then! Lets get the CSSWG to change that,

[astearns]

We switched from display-p3 to image-p3 in 2016, but I can’t actually find the discussion behind making the change 619c1fd

[jstblck]

Thank you for your reply @svgeesus.

The P3 colorspace was introduced into CSS Color at the request of Apple. We originally called it DCI P3 because that was what we thought it was. As more information emerged (different whitepoint, different viewing conditions and transfer curve) the naming was discussed in the CSS WG. image-p3 is the name that emerged as the favourite. Representatives from Apple were involved in those discussions.

I'm suprised Apple didn't want it to be called display-p3 instead of image-p3.

Adobe RGB (1998) is the official, trademarked, name of the Adobe colorspace. For licensing, trademark and copyright reasons, that phrase can't be used to describe the colorspace used in CSS. (I know). Nor can we use the phrase Adobe which would be a registered trademark infringement.

By the way there were several variants of the Adobe colorspace, so the 1998 specifier is an important part. This is why CSS has a colorspace called a98rgb.

It might indeed be more consistent to write a98-rgb and also prophoto-rgb which would also be easier to read.

It's a shame adobe can't be used. Would a-rgb-98 be better? That seems more consistent but also easy to read and wright.

I just think the built-in profiles need a naming scheme. Ideally, one that's consistent, easy-to-parse, and scaleable.

[css-meeting-bot]

The CSS Working Group just discussed Predefined colorspaces (renaming image-p3 to display-p3), and agreed to the following:

• RESOLVED: Take what we're calling image-p3 in our spec and rename it to display-p3

The full IRC log of that discussion

<dael> Topic: Predefined colorspaces (renaming image-p3 to display-p3)
<dael> github: https://github.com//issues/4056#issuecomment-505993145
<chris> I looked into the edit history, there were a number of changes drom dci-p3 to where we are now. Rik was the main contributor; there was discussion on the commits. See for example https://github.com//pull/557/commits/644c4c49b86d97bb59bf74e533bdfbe339f3d176
<TabAtkins> s/TabAtkins/tantek/
<dael> chris: Looked at htis. When originally put in spec confusion between dcip3 and what apple used. I originally called it dcip3 and that was wrong for many reasons
<dael> chris: Thought had resolution, but can't find. Went through edit history and it went through name changes. I'm happy changing to to what everyone knows it as
<dael> dino: My memory is not great. At some point I posted the profile for it given the idea it can be linked form the spec. I'm in favor of display-p3. Also willing to add another named to dci-p3 but I don't think there's a need
<dael> chris: I don't think there is. dci-p3 has strange greenish white and designed for use in digital cinema. Not appropriate for web
<dael> dino: That's the main driver for Apple creating what we call display-p3
<dael> tantek: You said because it's designed for dark env it's not appropriate for web
<dael> dino: Pitch black rooms
<dael> tantek: People do use the web in black rooms
<dael> dino: This is just hte convenience method. You can get the same result using display-p3. It just happens to have been calibrated for different env.
<dael> tantek: Is that calibration not useful when looking at web in the dark?
<dael> dino: If you're writing to be used in something like a cinema sure
<dael> astearns: THt's an argument to add dci-p3, but this is about what do we call the thing that we know is not dci-p3 and we want to support it
<dael> AmeliaBR: Adding sep profile can defer until demand. There is demand to match the thing Apple does. If the definition is match Apple then we should match the name unless there's trademark reasons
<tantek> s/black rooms/dark rooms
<tantek> s/just hte/just the
<dael> dino: Nothing apple specific about this other than we gave it a name. That's why I posted the file. There's nothing secret or protected about this
<dael> chris: Given that I would like to call for resolution
<dael> astearns: I could not find reason why we changed from display-p3 to image-p3. There's some discussion about how web isn't only for displayed, but not terribly compelling.
<dael> astearns: Prop: Take what we're calling image-p3 in our spec and rename it to display-p3
<dael> astearns: Objections?
<dael> RESOLVED: Take what we're calling image-p3 in our spec and rename it to display-p3
<dael> dino: Anything we can do in CSS to stop people from using their phone in a cinema?
<dael> astearns: Happy to follow the incubation process

[svgeesus]

In terms of consistent hyphenation: everyone writes srgb (well, they write sRGB) and s-rgb would be odd.

On the other hand prophotorgb is a bit long and hard to read, prophoto-rgb might be an improvement. And if we did that, then a98-rgb would also be reasonable.

Leaving issue open for further discussion on those.

[svgeesus]

I'm suprised Apple didn't want it to be called display-p3 instead of image-p3.

We were unable to find an actual resolution to use image-p3 and indeed, on the call today Apple expressed a preference for display-p3 as indeed did everyone else who had an opinion. So I just made that change.

[svgeesus]

I should also add the Apple definition of DisplayP3 to the spec ref database and then add that reference to CSS Color 4.

[svgeesus]

I added the AdobeRGB (1998) reference: e2dfded and tobie/specref@99a7d7d

[Crissov]

sRGB, ppRGB, aRGB?

[jstblck]

I'm glad that image-p3 will now be called display-p3.

In terms of consistent hyphenation: everyone writes srgb (well, they write sRGB) and s-rgb would be odd.

I don't think sRGB should be named s-rgb.

The current built-in profiles are: sRGB, Display P3, Adobe RGB (1998), ProPhoto RGB, and Rec 2020. These are the common names.

I suggest that we first replace the spaces with hyphens (eg srgb, display-p3, adobe-rgb (1998), prophoto-rgb, and rec-2020).

Then if necessary, I suggest we add a hyphen and the version to the end (eg adobe-rgb-1998 or adobe-rgb-98 [I'm not sure which]). Currently, this is only necessary for Adobe RGB.

Also if necessary, I suggest we shorten or replace trademarked names (eg adobe-rgb-1998/adobe-rgb-98 could be shorted to a-rgb-1998/a-rgb-98). Again, this is only necessary for Adobe RGB currently.

The naming scheme would then be the common name (unless trademarked and with hyphens instead of spaces), and if necessary, a hyphen and the version added to the end (such as the year for Adobe RGB).

PS: Is it a policy of the W3C that its standard don't include trademarks, or is it that Adobe would object to us using its name?

According to Adobe:

Legal note regarding color-space naming: Only the Adobe RGB (1998) ICC profile created by Adobe Systems Incorporated can accurately be referred to as "Adobe RGB (1998)." ICC profiles created by other vendors, even if they conform to the color image encoding described in the Adobe RGB (1998) color image encoding document, cannot be referred to as "Adobe RGB (1998)." If vendors choose to create their own profile according to this specification, and they want to indicate to their customers that this profile was written in accordance with Adobe's specification, then an alternate phrasing is required, such as "compatible with Adobe RGB (1998)."

Does that mean that Adobe would allow something like adobe-rgb-1998-compatible? I'm not suggesting we use that, but maybe they would allow it to be shortened to adobe-rgb-1998/adobe-rgb-98?

[tabatkins]

Summarizing the proposal:

• Add hyphens in the obvious places to make 'prophoto-rgb' and 'rec-2020'
  (consistent with 'display-p3')
• Add hyphens to 'a98rgb' in one of the following ways:
  • a-98-rgb
  • a-rgb-98
  • a-rgb-1998
  • adobe-rgb-98
  • adobe-rgb-1998

[svgeesus]

Or a98-rgb? (consistent with prophoto-rgb rather than pro-photo-rgb)?

PS: Is it a policy of the W3C that its standard don't include trademarks, or is it that Adobe would object to us using its name?

Both :)

[svgeesus]

Changed to display-p3, prophoto-rgb, a98-rgb and rec-2020

[tabatkins]

This works for me. You still want telcon discussion, Chris, or are you fine with this?

[Crissov]

I know, “Rec. 2020” is common jargon for ITU-R Recommendation BT.2020, and thus rec-2020 does make sense, but wouldnʼt bt-2020 or itu-2020 work just as well while being a bit more specific? 🤷‍♂️

[svgeesus]

I'm fine with this.

@Crissov as you say, rec.2020 is the widespread term of use. The full reference is in the references section, so there is no ambiguity.

Or we could rename it to rec-itu-r-bt-2020-2-2015 for the ultimate in self-descriptive identifiers.

[tabatkins]

We did briefly chat about this on telcon, just to confirm that it was indeed sufficiently addressed here in the thread and didn't need discussion. ^_^

@jstblck, can you confirm you're happy with the names Chris changed to?

[karip]

The rec-2020 name isn't anymore matching the rec2020 name in color gamut media query. It would be nice if the same name is used in both specs so it is easier to remember.

(on the other hand, media queries uses p3 instead of display-p3, so maybe this isn't a huge issue.)

[AmeliaBR]

@karip, good point, maybe open a separate issue, since it's less clear which set of tokens should be changed (these ones or the media query ones).

But we might have a compat issue now with changing the media query — it's supported in Chromium and WebKit.

[jstblck]

@jstblck, can you confirm you're happy with the names Chris changed to?

Sorry for the delayed response. I'm happy with the changes.

[svgeesus]

@karip wrote:

The rec-2020 name isn't anymore matching the rec2020 name in color gamut media query. It would be nice if the same name is used in both specs so it is easier to remember.

(on the other hand, media queries uses p3 instead of display-p3, so maybe this isn't a huge issue.)

Both true, and I think not a problem. Color 4 uses the terms rec-2020 and display-p3 to mean specific, objectively quantifiable things. MQ uses similar terms to mean loose, subjective things - in effect MQ has three buckets to describe the color gamut:

• barely good enough (sRGB) or even below
• wideish, but not super wide (p3)
• super wide (rec2020), mainly for upwards compatibility.

A display which covers 100% of "Adobe RGB (1998)" but only covers most of Display-p3, for example, would still fall into the "wideish gamut" bucket and match a p3 query.