Pages: [1]   Go Down

Author Topic: Learning process part II  (Read 5089 times)

NickXavi

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
Learning process part II
« on: February 09, 2016, 03:12:20 pm »

In my opinion, one of the most difficult part of learning to create printer profiles is as objectively evaluate the profiles.

I like Gamutvision by the precise drawing of the profile, the volume, roundtrip and Dmax.

I like drawing profiles provided by i1Profiler to see immediately mountains or unwanted valleys.

I like the impression of Colorchecker and comparative analysis provided by i1Profiler. Empirical analisis.

I like to see the Granger rainbow to immediately see subtle banding.

But to get an overview of the profile, the Oval gradient provided by Andrew Rodney in his book suits me very well.

Some images are attached as the result of softproofing the Oval gradient by profiles created for "Hahnemuhle Fine Art Baryta" paper, by Hahnemuhle (canned profile), by ColorMunki and i1Pro 2.

Some comments about these images come to my mind.

Of course there are other ways of evaluating printer profiles.

The most important, visual inspection of some test image. But sometimes I can see the differences and sometimes not.

I also searched a Windows version of the application "Alwan ColorPursuit", that Rodney cites in his book, but I have not found.

I would like to know opinions on this subject.

Thanks!




Logged
CG277, P800, i1Pro 2

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Learning process part II
« Reply #1 on: February 09, 2016, 03:29:31 pm »

AFAIK, Alwan ColorPursuit is no more.
The "Holey" checker that i1P uses is kind of cool, too bad when testing in their app, the don't tell you what rendering intent you're testing (and you're 'stuck' with one).
Synthetic images do tell you a lot about the smoothness of the various color engines. But real images are often a good way to test the profiles too. The Roman 16 suite is pretty awesome for testing output with real photography: http://www.roman16.com/en/
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Learning process part II
« Reply #2 on: February 09, 2016, 05:54:20 pm »

AFAIK, Alwan ColorPursuit is no more.
The "Holey" checker that i1P uses is kind of cool, too bad when testing in their app, the don't tell you what rendering intent you're testing (and you're 'stuck' with one).
Synthetic images do tell you a lot about the smoothness of the various color engines. But real images are often a good way to test the profiles too. The Roman 16 suite is pretty awesome for testing output with real photography: http://www.roman16.com/en/

The I1 "holey" colorchecker profile verification image is printed using Absolute Colorimetry.

The I1 Profiler has a set of downsized Roman 16 images in the user appdata folder.
Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Learning process part II
« Reply #3 on: February 09, 2016, 05:55:45 pm »

The I1 "holey" colorchecker profile verification image is printed using Absolute Colorimetry.
Well that sucks. Now how hard would it be for them to allow us to use other RI options? Don't answer. They haven't paid any love to the product in a long time.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Learning process part II
« Reply #4 on: February 09, 2016, 05:59:07 pm »

Well that sucks. Now how hard would it be for them to allow us to use other RI options? Don't answer. They haven't paid any love to the product in a long time.

Why on earth would one want to print a colorchecker using some other intent?  The only reason to print one is to check the profile accuracy by comparing it to an actual colorchecker and you can't do that printing RI or PI. If the AI matches then the RI is also good. PI,RI, SI are best tested with other images for smoothness and the colorchecker print is useless for that.
Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Learning process part II
« Reply #5 on: February 09, 2016, 06:00:18 pm »

Why on earth would one want to print a colorchecker using some other intent?
To test em all. We've been over this before.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Learning process part II
« Reply #6 on: February 10, 2016, 02:00:32 am »

To test em all. We've been over this before.

Exactly how would printing the colorchecker using an intent other than Absolute Colorimetric be of any use? How does one "test it?"  Is X-Rite expected to provide a set of holey colorcheckers with the dimmer patches that would be required to match these other intents? For that matter a color checker is a rather limited set of colors. It's main use is to verify a workflow against gross error. It's not useful for determining smoothness or out of gamut mapping.

The issue I have with X-Rite's color checker verification tool is that they don't publish their targeted colors.
Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Learning process part II
« Reply #7 on: February 10, 2016, 10:54:44 am »

Exactly how would printing the colorchecker using an intent other than Absolute Colorimetric be of any use? How does one "test it?"  Is X-Rite expected to provide a set of holey colorcheckers with the dimmer patches that would be required to match these other intents?
We've been over this time and time again. You have this idea that Abs Colorimetric is the right answer all the time or has something to always do with accuracy. I want to see how all the tables behave. And I don't want to use Abs in most cases. No, X-rite doesn't have to provide a series of targets! They have to provide the option to test the various rendering intents when we output the data to examine with the same target provided. Simple.
Now I do have a question: Who at X-rite told you the RI is Abs Colorimetric in this module?
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Learning process part II
« Reply #8 on: February 10, 2016, 10:57:17 am »

The issue I have with X-Rite's color checker verification tool is that they don't publish their targeted colors.
They should all be the same, have been published (and you could measure your own) or, download the synthetic Lab produced Macbeth and look at the lab values. Or,

ColorChecker RGB Summaries, Spreadsheets and Lab TIFF File
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Learning process part II
« Reply #9 on: February 10, 2016, 01:51:16 pm »

They should all be the same, have been published (and you could measure your own) or, download the synthetic Lab produced Macbeth and look at the lab values. Or,

ColorChecker RGB Summaries, Spreadsheets and Lab TIFF File

No, they are not "All the same."

Colorcheckers have changed over the years with a major change in 2014 because of laws restricting some of the pigments used earlier. The changes are significant spectrally and even the D50 Lab values differ. Some of the dE2k differences even exceed 1 and the dE76 variations are even greater.

http://www.babelcolor.com/colorchecker-2.htm#CCP2_data

Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Learning process part II
« Reply #10 on: February 10, 2016, 02:12:19 pm »

No, they are not "All the same."
Colorcheckers have changed over the years with a major change in 2014 because of laws restricting some of the pigments used earlier.
Fine, measure your own. IF indeed they have differed, which is correct? I've got at least half a doze (large, small, Passport) and of course the one with the holes. And further, I see zero reason why I can't pick any RI for testing. Lastly, can you tell me how you know the RI in i1P uses Abs Colorimetric?
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

NickXavi

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
Re: Learning process part II
« Reply #11 on: February 10, 2016, 02:16:23 pm »

Hi Andrew, knew these images, but had not paid attention. I will try.

Hi Doug, very interesting is the presence of these downsized images in i1Profiler. Its folder is: iC:\Program Files (x86)\X-Rite\i1Profiler\preview-images. I will make a first test with them.

Colorchecker: I compared the measured prints of i1Profiler Colorchecker with a reference measurement file (.cmxf) containing the 30 chart average spectral measurements of BabelColor with not DeltaE 2000 bad results. See my post "Learning process" with the results.

In fact my results wiht the averaged data was best that my results with a measured data with i1Pro 2 of two different actual Colorchecker Classic charts.

Spectral data that I used (April 2012): http://www.babelcolor.com/index_htm_files/CC_Avg30_spectrum_CGATS.txt

Thanks!
« Last Edit: February 10, 2016, 02:35:23 pm by NickXavi »
Logged
CG277, P800, i1Pro 2

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Learning process part II
« Reply #12 on: February 10, 2016, 02:35:16 pm »

Fine, measure your own. IF indeed they have differed, which is correct? I've got at least half a doze (large, small, Passport) and of course the one with the holes. And further, I see zero reason why I can't pick any RI for testing. Lastly, can you tell me how you know the RI in i1P uses Abs Colorimetric?

Simple.  When the colorchecker image is printed to be viewed against the colorchecker proof (the one with holes), it removes the scaling to the paper's white point. If it did not do that, and the paper has a ton of OBs, the whole image would be color skewed because all the rendering intents with the single exception of Absolute adapt to the paper white.

In fact, because the paper's white point has an L value below 100 and all other intents are scaled to the paper white any other rendering intent will also decrease overall luminance.

Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Learning process part II
« Reply #13 on: February 10, 2016, 02:46:25 pm »

Simple.  When the colorchecker image is printed to be viewed against the colorchecker proof (the one with holes), it removes the scaling to the paper's white point. If it did not do that, and the paper has a ton of OBs, the whole image would be color skewed because all the rendering intents with the single exception of Absolute adapt to the paper white.
And I WANT TO SEE THAT!
There's zero reason I can see that the RI can't be selected by the user. It simply provides options and new data points. Of course, if your goal is to force a 'better' match to make the product 'appear better' the current behavior makes a lot of sense from a marketing perspective.
I'm curious who told you that the RI in this module is Absolute Colorimetric.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Learning process part II
« Reply #14 on: February 10, 2016, 03:54:41 pm »

And I WANT TO SEE THAT!
There's zero reason I can see that the RI can't be selected by the user. It simply provides options and new data points.
The reason is clear. The purpose of printing to match a colorchecker proof is to match the proof. Absolute Colorimetric Intent IS what ones uses to match colors. Using any other Intent will not produce a match and intentionally so. This would just confuse people trying to make their own profiles. The patch luminosity will be reduced and the colors shifted to the paper white point. Your desire for additional "data points," is nonsensical.
Quote
Of course, if your goal is to force a 'better' match to make the product 'appear better' the current behavior makes a lot of sense from a marketing perspective.
It seems X-Rite's goal is obvious. To make it as easy as possible for people to see if their profile generation workflow and spectral measurements were w/o some significant error.

Quote
I'm curious who told you that the RI in this module is Absolute Colorimetric.

Who told you 2+2=4?   This is not rocket science and I'm frankly shocked it isn't obvious to you that I1P prints the proof image in Absolute Colorimetric Intent.

But that aside if you want to "SEE THAT" do this simple experiment:

1. Grab a profile from a paper with high OBs with the spectral data embedded. Plop it into I1P.
2. Create a new profile from that data.
3. Print the CC proof image. This is required since I1P doesn't let you print a proof image unless you just created a profile.
4. Now put the CC proof template over it, in normal (minimal uV) light you will see the printed gray patches have a nice yellow tint. Now look at it outside with the template and they will look neutral.

This is because daylight has close to the uV light of D50 and the profile was made with M1 illuminant which is based on the same. Absolute Colorimetric Intent adjusts the colors by adding a bit of yellow so that when viewed in daylight the fluorescence shifts that yellow back to a neutral color.

All other Rendering Intents do not and the CC proof would appear bluish not neutral in the same light.


Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Learning process part II
« Reply #15 on: February 10, 2016, 04:03:25 pm »

The reason is clear. The purpose of printing to match a colorchecker proof is to match the proof. Absolute Colorimetric Intent IS what ones uses to match colors.
That may be your reason... again, I want to see what each RI produces. It's useful info. But I can't.
Quote
Using any other Intent will not produce a match and intentionally so.
Might, might not. Doesn't change my need to see what each table does using this process.
Quote
This would just confuse people trying to make their own profiles.
That doesn't wash! It might confuse you.  :o 
What's confusing is we're not told what's going on. Given options, even with a default dropdown set to your beloved Abs Colorimetric, no harm done. No confusion.
Quote
The patch luminosity will be reduced and the colors shifted to the paper white point.
It might, it might not, I'd like to see the results if you don't mind.
Quote
Your desire for additional "data points," is nonsensical.
Thanks!
Quote
It seems X-Rite's goal is obvious.
Where would one find that fact from X-rite?
Quote
This is not rocket science and I'm frankly shocked it isn't obvious to you that I1P prints the proof image in Absolute Colorimetric Intent.
So you're assuming, kind of what I thought after you didn't actually reply to a specific question asked two times.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Learning process part II
« Reply #16 on: February 10, 2016, 04:25:31 pm »

So you're assuming, kind of what I thought after you didn't actually reply to a specific question asked two times.

No. I'm not assuming anything. I know they are using Absolute Colorimetric Intent when printing the CC proof.

Which part of this do you not understand?

If you want to why X-Rite MUST be using Absolute Colorimetric Intent do this simple experiment:

1. Grab a profile from a paper with high OBs with the spectral data embedded. Plop it into I1P.
2. Create a new profile from that data.
3. Print the CC proof image. This is required since I1P doesn't let you print a proof image unless you just created a profile.
4. Now put the CC proof template over it, in normal (minimal uV) light you will see the printed gray patches have a nice yellow tint. Now look at it outside with the template and they will look neutral.

This is because daylight has close to the uV light of D50 and the profile was made with M1 illuminant which is based on the same. Absolute Colorimetric Intent adjusts the colors by adding a bit of yellow so that when viewed in daylight the fluorescence shifts that yellow back to a neutral color.

All other Rendering Intents do not and the CC proof would appear bluish not neutral in the same light.


Logged

digitaldog

  • Sr. Member
  • ****
  • Online Online
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Learning process part II
« Reply #17 on: February 10, 2016, 04:27:39 pm »

Which part of this do you not understand?
Mostly the part where you tell me the features I want and can't have are OK with me (they are not).
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Learning process part II
« Reply #18 on: February 10, 2016, 04:40:49 pm »

Mostly the part where you tell me the features I want and can't have are OK with me (they are not).

I'm not telling you what options of features you want. I am saying that providing anything other than Absolute Colorimetric intent when printing a CC proof produces a less colorimetrically accurate proof. You are free to want to do whatever you like but the alternatives are less colorimetrically accurate and lower in luminance. It is not a question of maybe.

No idea where you got the idea I love AI.  I don't normally use Absolute Intent. Or rather, I never use it for normal printing. I do use if for proofing as is the norm and I do use if for reproduction work when it is desired that the reproduction white point match the original. Those are what Absolute Colorimetric Intent is designed for.
Logged

luxborealis

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2798
    • luxBorealis.com - photography by Terry McDonald
Re: Learning process part II
« Reply #19 on: February 14, 2016, 04:43:59 pm »

While my point of view is contrary to what many believe, have a read of this post...
Logged
Terry McDonald - luxBorealis.com
Pages: [1]   Go Up