Color management myths and misinformation video

[Lundberg02]

OK, now that we have decided that the number ofangels dancing on the head of a pin is between a positive integer n and infinity, how about these myths:
That there is any design standard for a monitor to be defined as aRGB or sRGB or that there is a defining organization that could create one.
That there is any agreement on the number of colors a human  can discriminate, or even what "discriminate" means.
That anyone knows what ColorSync is doing or why it lies to you.
That Eizo aRGB is the same as Dell aRGB, or any two manufacturers compared.
That any output image or video can be said to be in aRGB or sRGB after color grading.

I'm sure there are more but I'm not awake enough right now.

[Jim Kasson]

There is no difference in 'structure' in XYZ and any other color 'space' derived from it by a 3x3 matrix. <snip> They are not different spaces. They are what may be considered different coordinate systems within the same 3D space. 

Dr. Joofa,

You made that point in another context a month or two ago, and it resonated with me. Upon reflection, I think your use of the word space here is mathematically accurate, although it is considerably different from the usual usage in the color management world. I don't propose that the color management world change its terminology -- I think that would be an uphill struggle to win the hearts and minds of a ship that's already sailed (Excuse me, I love mixing metaphors). Maybe you could propose another word that captures your meaning?

On a related point, I can see that any reversible  linear transformation (unlike XYZ to xy, which is linear, but is a projection of a 3D space on a 2D one, so is a one-way trip) of a method conversion of continuous spectra to 3D values -- defined by a set of color matching functions -- can be considered mathematically to be different ways of looking at the same thing. In your mind, is that also true if each plane is subjected to a continuous nonlinearity which possesses a continuous inverse? That would cover colorimetric RGB color spaces. It also, with a little finessing, cover CIEL*u*v*.

Jim

[joofa]

You made that point in another context a month or two ago, and it resonated with me. Upon reflection, I think your use of the word space here is mathematically accurate, although it is considerably different from the usual usage in the color management world. I don't propose that the color management world change its terminology -- I think that would be an uphill struggle to win the hearts and minds of a ship that's already sailed (Excuse me, I love mixing metaphors). Maybe you could propose another word that captures your meaning?

I have no problem with saying, say RGB and XYZ, are color spaces, as long as it is understood that they coexist in a certain sense as depicted in the diagram below (CIE RGB and XYZ):

If this geometry is not realized then people make errors in color calculations.

For e.g., the matrix for conversion from CIE RGB->XYZ is:

          0.48872   0.31068   0.20060
     M =  0.17620   0.81298   0.01081
          0.00000   0.01020   0.98980

Now take the saturated Red = [1 0 0]'. The vector length in RGB space is 1. However, the same color is represented in XYZ as M*[1 0 0]', and its vector length is now 0.52. Why the difference? Because, the correct geometry, as illustrated by the above diagram was not used. If that geometry is used then the vector length is the same.

Such things have resorted people to jump too quickly to non-linear analyses (non-linear spaces), using all kinds of strange profile measures, etc, in certain domains.

[digitaldog]

Getting ready to record part one of my video. I've got an option to record on the Wide Gamut NEC or as I've always done on the past (to be safe), a MacBook display which is "sRGB-ish"

The differences in the output of the Gamut Test Print don't translate that well to video but none the less I do see some useful differences. I'm was wondering if the application I use (ScreenFlow) would handle the wide gamut display properly so I did a very quick test showing one freeze frame of the prints side by side (viewed in LR's Compare View). I'm wondering if a few of you can tell me the two versions look OK and as I see them in Safari, basically the same. I've got about 5 seconds of each option. Unfortunately the scaling is a tad different. But it appears to me there is no difference if I record the presentation on the NEC or the MacBook display but I'd like to know if other's see the same thing.

Here's the clip: http://www.digitaldog.net/files/DisplayTest.mov

Thanks!

[Steve Upton]

Getting ready to record part one of my video. I've got an option to record on the Wide Gamut NEC or as I've always done on the past (to be safe), a MacBook display which is "sRGB-ish"

The differences in the output of the Gamut Test Print don't translate that well to video but none the less I do see some useful differences. I'm was wondering if the application I use (ScreenFlow) would handle the wide gamut display properly so I did a very quick test showing one freeze frame of the prints side by side (viewed in LR's Compare View). I'm wondering if a few of you can tell me the two versions look OK and as I see them in Safari, basically the same. I've got about 5 seconds of each option. Unfortunately the scaling is a tad different. But it appears to me there is no difference if I record the presentation on the NEC or the MacBook display but I'd like to know if other's see the same thing.

i don't think I can see any difference myself. Can you see a difference on your display? (I'm viewing it on an iMac display so outer gamut diffs may not show up... or would they...?)

Also, do you know where you'll be posting your video when it's done? (as in URL). We're doing a magazine article on Color Management Myths (different industry) and it might be appropriate to put a link into it, depending on the content and focus of the video...

Steve Upton

[digitaldog]

i don't think I can see any difference myself. Can you see a difference on your display? (I'm viewing it on an iMac display so outer gamut diffs may not show up... or would they...?)

No, I see no difference on either and using Safari which is kind of good (it means ScreenFlow is color managed). But I wanted to be on the safe side before moving towards recording.

Also, do you know where you'll be posting your video when it's done? (as in URL).

Low Rez will be on YouTube, high rez on my site. Thanks.

Update: Oh, as to when done, not sure. I'd like to get it done by next Tuesday after which I'm out of town working till the following Sunday. I'll let everyone know when it's done. Not sure if it will be one big fat, long video or two (perhaps three). It's going to be three parts but part 1&2 might be one video and part three (the geeky part) optional. Have to see how it goes in editing.

[digitaldog]

OK, got a rough cut here. Video is pretty high rez at this point.

http://digitaldog.net/files/WideGamutPrintVideo.mov

Leaving early tomorrow for a job and while I'll have the footage and editing software on my MacBook, not sure if I'll have time to do any further editing or 'fixes' (if necessary). Doesn't mean I'm not done polishing or ignoring any comments here, just engaged in 'real work' until early nest week  .

[TonyW]

Like what you have done with this although not seen all - need a rest before I view and try to understand part III 

I do however have a question relating to the Printing the Gamut Test File - it may be a stupid question but... 

I noticed that you had Send 16 bit Data option ticked.  As you know this option not available for Windows users. 

I believe that 16 bit print options have been discussed at length here and elsewhere on 'the net' and my understanding is that most seem to see little if any advantage. 

I have not had the opportunity to print your test file as yet but wondering if you would expect differences particularly of the synthetic images to be more/less noticeable or the same in an 8 bit print pipeline?

[digitaldog]

I noticed that you had Send 16 bit Data option ticked.  As you know this option not available for Windows users. 

Right but it will not make any difference I've ever seen on the print. And on Windows, it will still be apples to apples comparison. I don't think you'll see any difference without that setting. I expect some viewers may not be using Epson printers at all

[bjanes]

OK, got a rough cut here. Video is pretty high rez at this point.

http://digitaldog.net/files/WideGamutPrintVideo.mov

Leaving early tomorrow for a job and while I'll have the footage and editing software on my MacBook, not sure if I'll have time to do any further editing or 'fixes' (if necessary). Doesn't mean I'm not done polishing or ignoring any comments here, just engaged in 'real work' until early nest week  .

Andrew,

I downloaded your test image and printed the images according to your instructions using my Epson 3880 with the Epson profiles on Premium Glossy paper. Most of your observations were confirmed, but the results on Bill Atkinson's balls were different and unexpected with the blue images. The results were reasonably well predicted by softproofing and the image below represents my printed images very well. The blue balls show banding on the higher luminance ball and the lower luminance ball is very dark. The Epson profiles are apparently not as good as your custom profile. What do you think? The file is AdobeRGB, which is necessary to represent the gamut of my NEC PA241W.

Bill

[digitaldog]

Not sure Bill but when I get home, maybe I can dig up some glossy paper and see what the Canned profile does on this end.

[Tim Lookingbill]

OK, got a rough cut here. Video is pretty high rez at this point.

http://digitaldog.net/files/WideGamutPrintVideo.mov

Leaving early tomorrow for a job and while I'll have the footage and editing software on my MacBook, not sure if I'll have time to do any further editing or 'fixes' (if necessary). Doesn't mean I'm not done polishing or ignoring any comments here, just engaged in 'real work' until early nest week  .

You really nailed the presentation on the video, Andrew. Good work.

I'm seeing the differences viewing on my "sRGB-ish" 27" LG display. Just some nit-picky suggestions on what I'm seeing that you might want to emphasize...

1. There is a marked difference in saturation and hue in the gold color in the Granger Rainbow at the "blue shorts guy in the boat" section that you skipped over.

2. There's quite a bit of video compression induced banding on a few Granger Rainbow comparisons (not all) that make it hard to distinguish whether it's on the print or in the video.

All in all you've made a very convincing argument for editing in ProPhotoRGB which I've always adopted since I began shooting and processing in Raw.

Now I'm going to see if these differences show up printing on my $50 Epson NX330 "All In One" with "Printer Manages Color" setting just out of curiosity. It might explain some odd color transitions I sometimes get printing on Epson Ultra Premium Glossy paper.

[digitaldog]

2. There's quite a bit of video compression induced banding on a few Granger Rainbow comparisons (not all) that make it hard to distinguish whether it's on the print or in the video.

I can play around a bit more in exporting the video, I'm not real savvy on all the permutations in ScreenFlow for video output. I can make it a higher rez (which will make for a much bigger download). It will look even worse on YouTube!

[Lundberg02]

Please correct the spelling of "benefits" in the title page!
Also after saving it to the desktop as a webarchive, it will not load and play. I saved it a couple other ways and will try them.
I have always printed from ProPhoto so I am very glad to see this now that I have a wide gamut monitor, thanks for your good work.

[digitaldog]

Please correct the spelling of "benefits" in the title page!
Also after saving it to the desktop as a webarchive, it will not load and play. I saved it a couple other ways and will try them.

Good catch, fixed. Thanks. Not sure why the video doesn't just play through your web browser. It's a .mov and runs fine on this end (Mac).

[jed best]

Hi Andrew,

Very nice video. I use Eizo monitors and can clearly see the differences between ProPhoto and sRGB. The only color that does not show a significant difference on my monitors is yellow. Hope this feedback helps.


Jed

[Lundberg02]

The video plays fine if I just save the URL in a document.
 Odd that you don't see improvement in yellow. That's one of the reputed advantages of Pro Photo.

[Schewe]

Odd that you don't see improvement in yellow. That's one of the reputed advantages of Pro Photo.

Actually, it's red/orange that PFRGB provides benefits...

[Tim Lookingbill]

This could be a display gamut issue with what Lundberg and I are seeing in the marked difference in yellow between sRGB/PFrgb in the video. Or it could be better hardware LUT calibration/software/profile on jedbest's Eizo. Not sure.

I'm on a $330 27" sRGB-ish gamut display calibrated with a Colormunki Display.

[Tim Lookingbill]

This is a screenshot of what I see creating an intense yellow in Photoshop's Color Picker with no clipping in ProPhotoRGB and convert to sRGB on a duplicate.

The file below has my display profile embedded, no converting to sRGB. You should see a difference because the RGB numbers are different.