V2 profiles, V4 profiles, and RT's Out-of-gamut indicator

(Elle Stone) #2

Without having the profiles and the image (or a reduced size version) to look at, and also without the RT parameters, it’s hard to even guess what’s going on.

If you have the actual AdobeRGB1998 profile made and distributed by Adobe, you can’t share just the profile. But try this: substitute “ClayRGB1998.icm” from the ArgyllCMS “ref” folder for your version of AdobeRGB and see if the same situation you describe still happens.

(Andrew) #3

Thanks Elle. I was just getting some files together to post up, and I now reckon this is a bug which happens the first time through with a freshly-invoked RT. To reproduce, before I forget -

  1. Run RT (5.4.924 built 11 Oct. is mine)
  2. Into editor CR2
  3. Change monitor profile to RTv2_Medium and click out-of-gamut button, see cyan areas
  4. Save as jpeg
  5. Click out-of-gamut button, i.e. switch off cyan
  6. Click out-of-gamut button, cyan should re-appear, but it doesn’t.
  7. Go back to browser, repeat, but no cyan, have to quit RT and restart to see it again.
    @Morgan_Hardwood, it’s a bug!

So first time thru, v2 gives cyan but it then goes away. But first time thru with v4, there is no cyan. So in addition, there seems to be some difference between these two profiles involved in the bug.

@Elle, I’ll maybe look at the Clay version when I get time. My point was that there was a big difference between v2 and v4 which I didn’t understand.

(Elle Stone) #4

The Clay profile is a V2 profile that is an exact match other than the name to the AdobeRGB1998 profile distributed by Adobe.

I can think of a couple reasons why you might get a difference between different versions of “the same” profile when soft proofing, depending on this and that including what sort of TRC is being used by the different profiles. But without the actual image and profiles to look at it’s not easy to figure out what’s actually happening.

(Andrew) #5

Let’s first see what the devs say. I’m guessing it’s a bug which when fixed will mean little or no difference in OOG appearance between the various “adobeRGB” profiles. (Devs - I haven’t tried other monitor profiles re. this bug)


As I cannot reproduce now with a CR2 image, I have some questions :

  • what is your working profile
  • in your step 3, when you say “monitor profile”, I suppose it is the
    “output profile”. Or do you go to preferences to effectively change the monitor profile?
  • In the last case I don’t know if V4 profiles are supported in my system (W10) as display profiles.

(Andrew) #7

hi, thanks for looking into this.
Working profile is Rec2020.

Output profile? - No! - I am definitely talking about the Monitor profile, which of course you can change in prefs but I am changing it using the dropdown at the bottom of the editor screen. You can see this in the .pngs above near the middle at the bottom. (Output profile is RTv2_Medium in case this is relevant)

V4 support. I’m not sure this matters? Surely we are saying to RT “Calculate what is OOG based on the colour space implied by profile XYZ and display as cyan”. So your system is kind of irrelevant?

(Andrew) #8

hi again, I just downloaded the Morgin NEF and it’s behaving just the same as my raw.
Morgin.NEF.pp3 (10.4 KB)
Morgin.NEF (9.0 MB)
Here’s what I get -

Hope this helps.

Edit: In fact not quite the same as before, the OOG cyan re-appears after saving. The output profile was AdobeRGB1998. So I changed it to RTv2_Medium and the previous behaviour was back, i.e. no cyan on re-enabling OOG after a save.

(Andrew) #9

@gaaned92, I think my last edit was perhaps confused and may suggest there is no problem.
There is a problem! with RT incorrectly showing the OOG when freshly invoked. The steps I gave initially are basically correct. I believe RT is initially showing the OOG for the sRGB profile. The output profile is also involved. So…
Invoke RT, edit Morgin, set monitor and output profiles to RTv2_Medium, click the OOG button, see the cyan which I believe is sRGB OOG and therefore incorrect. Save as jpeg. Repeatedly click the OOG button and now there is no cyan - which I suspect is correct.

Why do I think it’s showing the sRGB OOG initially? Because if you repeat the above but with RTv2_sRGB as the output profile (monitor profile still medium) then you get the same cyan pattern.
But this is also wrong, surely. The cyan should be related to the monitor profile, not the output profile, because the adjacent soft-proofing button is not pressed.

(Desmis) #10


I am working on this problem… but it seems a little difficult :slight_smile:

First I think, v4 are not suitable, I propose to show only files in the selector, with RTv2

For the other problem, I realize only for ICC profiles as

  • “AdobeRGB” and “sRGB” with illuminant D65, and TRC as 2.2 or sRGB
    an other code to elabore these profiles which must be the ones useful for the monitor

For example, I compare:
RTv2_Medium with AdobeRGB1998, all tags (except of course description, copyright,…) are strictly the same
Rtv2_sRGB with sRGB_Color_space_profile, all tags are strictly the same

For the others cases, the “ICC profile creator” incorporate in RT should, as now, solve the majority of iCC profile problems (icon near Preferences)
When, I think, I have finish this work… to morrow…or later, I will push a new branch with these modifications


(Andrew) #11

hi Jacques, that’s great you’re working on it.
If I understand the problem correctly, it’s hardly a show-stopper, so I wouldn’t worry too much about doing it “tomorrow”!
I’m afraid I don’t really understand your comments about the tags and illuminant, but that probably doesn’t matter!


@RawConvert I thought I understood that stuff (choice of display profile in RT, soft-proofing, OOG, working profile…) but after your post, my experiments let me puzzled.
I cannot figure out how the tooltips of these tools are coherent with behaviour.

As my knowledge is very small, I tried and failed to find a comprehensive and understandable explanation of these tooIs somewhere in rawpedia.
So I cannot really answer now your questions.

(Andrew) #13

I think the tooltips make sense in themselves, but the program is doing something different, or has started doing something different since the two buttons were introduced.

(Andrew) #14

@patdavid or @paperdigits, in my post above, I definitely clicked the Reply in gaaned92’s post but my post is not showing arrowed to him, so maybe he’s not alerted. Whereas gaaned92’s reply to me clearly shows this. Am I doing it wrong?

(Mica) #15

I am not going to screenshot now since I’m on mobile, but make sure tbr users icon and name are in the top left hand part of the reply pop up.

Edit: I see what you mean now, and I’m not sure. I’ll look into it.



1- I cannot really understand what the problems are. Could you elaborate and explain how to reproduce in RT in a github ticket please?

@jdc, @Hombre, @agriggio, @heckflosse, @sguyader and other devs

Sorry to say that but as it is written now in rawpedia, I cannot get usable information from rawpedia color management.

2- I think the next task is to update the rawpedia color management pages to be coherent with present RT behaviour , comprehensive (e.g. it describes in simplified english what is supposed to do, what is done inside RT and where in the pipeline…) and understandable by a noob like me.

All the color management tools should be described or pointed to applicable page.(OOG, soft proofing, outpu tprofile, display profile, working profile …)
I hope some proficient dev will volonteer. Otherwise I can try to do that with your help ( identification of applicable threads, answer to questions, proof reading…).
Not for 5.5

  1. Once done, new improvments and modifications can be studied.

Bug in the last few 5.4 devs
(Desmis) #17

I created a new branch “enhance_icc” for the problems mentioned above

For the ICC V4, I deleted the display of v4 files (RTv4) for ICC used for the screen (in Preferences “default profile monitor”) because I think there is incompatibility (of course to check later).
I think also, for this type of profile (monitor display) that ICCV4 bring nothing - at least for this type of profiles

I thought a lot about the color gamut problem (OOG) between AdobeRGB1998 and RTv2_Medium, effectivly there are differences in some cases.

Now with modifications, the behavior seems to me identical, the main modification is the replacement of the gamma sRGB (g=2.4 s=12.92) by a simple gamma = 2.2, but also by changing the way it is calculated (gamma 2.2).

I change the code inside the “ICC Profile Creator”, but it does not affect the interface.

Before this modification, all profiles furnished (RTv2 or RTv4) were all by default with a “sRGB gamma”. Now RTv2_Medium and RTv4_medium are with gamma=2.2

Of course you can change these values with the “ICC profile creator”.

More complete explanations will be provided shortly, in the issue that I will open (probably tomorrow).

I am open to any proposal to improve Wikipedia, but my low level of English will be a handicap…
Color Management is not easy, with a lot of empiricism and approximation (Ciecam, gamut, Munsell correction, Lab, ICC, etc.), but some progamming solutions work better than others, here again I can contribute if I am told what to do.


(Andrew) #18

@jdc, hi jacques, regarding your post above -
I don’t understand gammas and slopes properly, but are you sure it’s ok to start changing them?! I mean if a current profile has been built with 2.4 and 12.92, is there perhaps a good reason for that?

Also, much of the problem I noticed was that RT is behaving differently on “first use” (i.e. when it has just been launched) as compared with later on (i.e. after doing a Save). So maybe some storage is not initialised properly?! To me this seems like something quite different from altering profiles.

(Desmis) #19

I know there is no Rawpedia for this problem… but there is no problem to change “gamma”
With “ICC profile creator” you can change what you want : illuminant, gamma, primaries (with v4)
So you can build - if those furnished do not suit you - any ICC profile you want :slight_smile:

(Desmis) #20

I push a change in “dev” with a lot of (small) enhancements for


(Andrew) #21

Thanks jacques, I will try and build it and play over the next few days.