Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: ICC profile "wtpt" tag; V2 vs V4 profiles and X-rite vs Colorthink  (Read 2879 times)
tvalleau
Newbie
*
Offline Offline

Posts: 29


« on: February 22, 2013, 06:59:25 PM »
ReplyReply

(did search here before posting this.)

Here's what's so: The ICC specs (ICCv43_2010-12, pages 23 and 40) for V4 (Version Four) monitor profiles -requires- that the "wtpt" tag contain the Lab and temp information for D50, regardless of the actual target temp (usually D65) of the profile. (The "chad" tag is used to calculate the actual temp.) Thus, when you look at the "wtpt" tag in ColorThink, on your > D65 < monitor profile, it will say "5000K."

This requirement was not so in V2 icc monitor profiles, and so the convention became to insert the actual target temp & Lab in the "wtpt" tag. A V2 "wtpt" tag will show "6500K" for a D65 profile and "5000K" for a D50.

However, it is not "illegal" for the V4 requirements to be used in a V2 profile, since it was never specified... it's merely convention.

So: here's the situation - V TWO profiles created by Apple, and Datacolor show "6500K" and V TWO profiles created by X-rite show "5000K".

It appears, at the very least, however, that ColorThink uses the temp as reported in the "wtpt" tag when graphing the spaces. That leads to the situation shown in the attachment, below.

In that image, BOTH are V2 6500K monitor profiles, one with "wtpt" 6500K (as convention dictates) and the other with "wtpt" 5000K (as X-Rite chooses.)

As you can see, the white points are off by 1500K and the comparison fails.

I'm merely reporting this. I do not know what, if any, other problems this confusion of "wtpt" temps may cause outside of ColorThink.

Thoughts would be appreciated.

OH: I  have thoroughly discussed this with X-Rite technical support, and I did suggest that they follow convention when creating V2 profiles. I've written to TechSupport at Chromix, and am awaiting a reply.

« Last Edit: February 22, 2013, 07:02:55 PM by tvalleau » Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9188



WWW
« Reply #1 on: February 22, 2013, 07:15:14 PM »
ReplyReply

Thoughts would be appreciated.

My thoughts are, V4 ICC anything is kind of a joke. It's poorly supported by many applications, doesn't bring much if anything to the party. It hasn't been accepted well or adopted well by those companies that even create them. Best to avoid them.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
RichWagner
Newbie
*
Offline Offline

Posts: 26


WWW
« Reply #2 on: February 22, 2013, 08:11:12 PM »
ReplyReply

My thoughts are, V4 ICC anything is kind of a joke. It's poorly supported by many applications, doesn't bring much if anything to the party. It hasn't been accepted well or adopted well by those companies that even create them. Best to avoid them.

I think you missed the problem, Andrew. Regardless of the merits/potential problems with v4 profiles, "avoiding them" does not address the problem reported here, where X-Rite Version TWO profiles are not following v2 tag conventions and instead use v4 conventions.  There are no v4 profiles involved in this problem, so avoiding them can't possibly fix the problem. Perhaps getting X-Rite to "avoid using v4" tag conventions when building v2 profiles would.

--Rich
Logged
tvalleau
Newbie
*
Offline Offline

Posts: 29


« Reply #3 on: February 23, 2013, 12:26:30 AM »
ReplyReply

Correct, Rich.

I've attached a screen grab of a Version 2 D65 ICC profile created by iProfiler, and how it looks in ColorThink. If the file doesn't show up, you can also find it here: http://cl.ly/image/263o3X2E323v

Logged
RichWagner
Newbie
*
Offline Offline

Posts: 26


WWW
« Reply #4 on: February 23, 2013, 01:10:41 AM »
ReplyReply


Here's what's so: The ICC specs (ICCv43_2010-12, pages 23 and 40) for V4 (Version Four) monitor profiles -requires- that the "wtpt" tag contain the Lab and temp information for D50, regardless of the actual target temp (usually D65) of the profile. (The "chad" tag is used to calculate the actual temp.) Thus, when you look at the "wtpt" tag in ColorThink, on your > D65 < monitor profile, it will say "5000K."

This requirement was not so in V2 icc monitor profiles, and so the convention became to insert the actual target temp & Lab in the "wtpt" tag. A V2 "wtpt" tag will show "6500K" for a D65 profile and "5000K" for a D50.

However, it is not "illegal" for the V4 requirements to be used in a V2 profile, since it was never specified... it's merely convention.

i'm not sure the v2 case was so simple... for example, see:

http://en.usenet.digipedia.org/thread/12764/10489/

I would be most interested to hear Steve Upton's explanation of why the 3-D gamut plots look different.  I suspect that the answer lies in the assumptions that are made when calculating the gamut plots of profiles and the color space used.  I vaguely remember something about this... but no time to research it tonight.  I suspect all of these profiles are valid profiles, and that the issue is just your ability to compare them in ColorTHink. That's not really X-rite's problem.

Can you post two of the profiles that you're comparing, and give the version of ColorThink(Pro?) that you're using?

--Rich
Logged
RichWagner
Newbie
*
Offline Offline

Posts: 26


WWW
« Reply #5 on: February 23, 2013, 01:25:13 AM »
ReplyReply

I would be most interested to hear Steve Upton's explanation of why the 3-D gamut plots look different.  I suspect that the answer lies in the assumptions that are made when calculating the gamut plots of profiles and the color space used.  I vaguely remember something about this... but no time to research it tonight.  I suspect all of these profiles are valid profiles, and that the issue is just your ability to compare them in ColorTHink. That's not really X-rite's problem.

Here you go.

See:
http://lists.apple.com/archives/colorsync-users/2005/Jul/msg00164.html

--Rich
Logged
tvalleau
Newbie
*
Offline Offline

Posts: 29


« Reply #6 on: February 23, 2013, 02:02:07 AM »
ReplyReply

Sure: they are posted below.

The version of ColorThink is 2.3 (not Pro.)

They are all Version 2 profiles, set to D65.
The one named "DELL(lotsOfNumbers)" was created by the Mac OSX 10.8.2 automatically.
The one named "testSpyder..." was created by Spyder 3 elite version 4.0.7.
The one named "fullFrontal" was created by iProfiler version 1.4.2.

The issue should be simple to test on one's own however: just create a V2 profile with iProfiler at D65, and compare it to (Map the spaces) against almost any other V2 monitor profile. If you have a Mac, grab Apple's default profile. If you have a Spyder, make a V2, D65.

The issue here is NOT "what is correct" since determining that isn't really possible. Only consensus is possible.

The issue is that because of the conflicting use of the "wtpt" tag in V2 profiles, it is not possible to accurately compare them using ColorThink... -and- that there may be other issues of which I'm not aware for the same reason.

One of my main uses of ColorThink (which I love) is to compare profiles. However, if I cannot visually compare two D65 profiles... um... you see my point, I assume.

The three profiles are zipped up below.
Logged
tvalleau
Newbie
*
Offline Offline

Posts: 29


« Reply #7 on: February 23, 2013, 02:04:19 AM »
ReplyReply

Thanks. Unfortunately, I'm old, and it's late where I am, so I'll read that tomorrow.

(I see I stumbled across something that may be an 8 year old issue, eh?  Doesn't bode well for its resolution, eh?)  :-)

Thanks again.

Logged
RichWagner
Newbie
*
Offline Offline

Posts: 26


WWW
« Reply #8 on: February 23, 2013, 06:18:40 AM »
ReplyReply

(I see I stumbled across something that may be an 8 year old issue, eh?  Doesn't bode well for its resolution, eh?)  :-)

See the attached screenshot made with ColorThink Pro 3.03.

--Rich
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9188



WWW
« Reply #9 on: February 23, 2013, 09:25:21 AM »
ReplyReply

(I see I stumbled across something that may be an 8 year old issue, eh?  Doesn't bode well for its resolution, eh?)  :-)

http://www.colorwiki.com/wiki/Color_Management_Myths_26-28#Myth_26
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
tvalleau
Newbie
*
Offline Offline

Posts: 29


« Reply #10 on: February 23, 2013, 12:30:29 PM »
ReplyReply

Well...-that- is interesting.  The latest version of ColorThink standard displays different comparison graph than ColorThink Pro.

I'm not so sure that's intentional. I'll pass it along.

Thank you.

Logged
tvalleau
Newbie
*
Offline Offline

Posts: 29


« Reply #11 on: February 23, 2013, 12:45:59 PM »
ReplyReply


Thanks. That's an explanation of ColorSync vs ColorThink, but it still does not explain why ColorThink gives two different results.

And, apparently from the screen shot taken of the results of comparing the same profiles in ColorThink PRO, show that the issue (of matching the monitor white points [note that monitors are treated uniquely by the ICC specs]) was addressed in the current Pro version, but remains unaddressed in the current standard version.
Logged
RichWagner
Newbie
*
Offline Offline

Posts: 26


WWW
« Reply #12 on: February 23, 2013, 09:42:42 PM »
ReplyReply

And, apparently from the screen shot taken of the results of comparing the same profiles in ColorThink PRO, show that the issue (of matching the monitor white points [note that monitors are treated uniquely by the ICC specs]) was addressed in the current Pro version, but remains unaddressed in the current standard version.

I think you've got the picture.  It is apparently on the "to do" list for ColorThink (standard) x years.  Do read and understand the rationale for the change in how profile gamuts are plotted, though.  The change is not so much a "bug" as it is a design decision that has changed.

http://www.colorforums.com/viewtopic.php?t=286&start=0&postdays=0&postorder=asc

You can ping Steve Upton and see when/if he plans to update the 3D graphing in CT standard, or, since you really seem to be into the nuts and bolts of color management and profiles, I'd highly recommend that you upgrade to CTPro.  I can't imagine using a neutered version of ColorThink.  It's really a go-to tool.  The worksheets, gamut volume, color lists, etc. are all tremendously helpful.  Once you've used them, you won't want to be without them.  And no, I get nothing from Chromix...

--Rich

Logged
tvalleau
Newbie
*
Offline Offline

Posts: 29


« Reply #13 on: February 23, 2013, 10:00:39 PM »
ReplyReply

I think you've got the picture.  It is apparently on the "to do" list for ColorThink (standard) x years.  Do read and understand the rationale for the change in how profile gamuts are plotted, though.  The change is not so much a "bug" as it is a design decision that has changed.

http://www.colorforums.com/viewtopic.php?t=286&start=0&postdays=0&postorder=asc

You can ping Steve Upton and see when/if he plans to update the 3D graphing in CT standard, or, since you really seem to be into the nuts and bolts of color management and profiles, I'd highly recommend that you upgrade to CTPro.  I can't imagine using a neutered version of ColorThink.  It's really a go-to tool.  The worksheets, gamut volume, color lists, etc. are all tremendously helpful.  Once you've used them, you won't want to be without them.  And no, I get nothing from Chromix...

Nice find on that link, Rich; Thanks! That does indeed seem to be the issue, and in a followup letter to Steve (sent before seeing your post), I posited as much.

It modestly rewarding that I seem to have discovered and then grasped the issue on my own. It certainly got me exploring intricacies I'd not bothered with before, and I learned a lot, due in no small part to the contributions in this thread by others.

If I learn anything of interest when Steve replies, I'll round out the thread by posting it here (with his permission, of course.)

My thanks again to one and all.
Logged
hueman
Newbie
*
Offline Offline

Posts: 8


« Reply #14 on: February 27, 2013, 02:23:33 AM »
ReplyReply


Wow, I had to reactivate my account here on LL. It seems the account I set up 11 years ago (! literally March 2002) became inactive along the way.

Anyway, thanks for your patience in waiting for my response to the thoughtful and remarkably accurate discussion on ColorThink graphing and white points.

I suppose most of the information has already been covered so I'll just work to confirm the statements already made and supply some history.

When designing the gamut graphing in ColorThink I really didn't have anything to follow. At the time there were no other gamut graphing tools except a very simple (yet inspiring) color plotting tool in ColorBlind. I made the decision to follow the "path of the pixel" in an ICC workflow through the Lab PCS and also graph gamut volumes in "absolute" terms. This meant that you could graph an ICC profile's printer gamut along with the measurements that created it and they would match up - assuming it was a good quality profile.

Strangely enough I think that ColorThink is still the only tool that does things this way.

Anyway, if printer profiles were to be graphed in Absolute terms - primarily meaning that the white point of the gamut is set to the actual Lab values of the paper color - then it seemed logical that monitor profiles should be as well. As color space profiles like sRGB and Adobe RGB are also ICC monitor profiles, they are treated the same way. This means that the white point recorded in display profiles is used regardless of the confusion and ambiguity that placed it there.

Mapping the white points to D50 in V4 profiles makes sense. It is something that was already done for most print profiles but not so for monitor profiles. I may have, in fact, contributed to the problem long ago when early versions of GretagMacbeth's monitor profiling software didn't write the calibrated white point of the display into the wtpt of the profile. I complained to GM about it and a later version showed up that did.... I'm not sure if my feedback caused the change but now that I've learned more about it I regret the original complaint. My problem was that I wanted some sort of "documentation" in the profile to record the calibrated white point of the display. I've since learned that the viewing conditions tag (view) is probably better place to record that information, though I don't know if any display calibration software does that. Also, I might not be 100% correct on all of this, it's been a while.

Back to ColorThink.

Over time I came to realize that, while I could argue that absolute gamut graphs for displays and working spaces was correct in one sense, comparing profiles with different white points just didn't work. I realized that when we view displays we adapt to their white points. If we had two displays calibrated to different white points and we looked back and forth between them, we would adapt to each in turn. So it makes sense to just adapt display profiles to the same D50 white point and view them together.

This decision came about at the time I was writing ColorThink Pro. As a result Pro adapts display profiles to D50 before graphing. ColorThink 2 is supposed to catch up but as you can tell, it hasn't yet. As we expected, most users who are analyzing color at this level have moved to Pro but this motivates me to revisit v2 and modify this behavior as well.

I hope this clears up, or at least confirms things. If you have any other questions or issues, fire away!

regards,

Steve Upton
Logged
tvalleau
Newbie
*
Offline Offline

Posts: 29


« Reply #15 on: February 27, 2013, 12:09:31 PM »
ReplyReply

Thank you for taking the time to respond here, Steve. In fact, everyone I've dealt with, from X-Rite, thru LL and Chromix has been courteous and helpful.

... and, FWIW, I'm now the proud owner of ColorThink Pro.

Thanks again to one and all.

Tracy
Logged
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad