Pages: [1]   Go Down

Author Topic: Argyll chartread test: please try to replicate (FIXED in v1.6.1)  (Read 12953 times)

NeroMetalliko

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 78
Argyll chartread test: please try to replicate (FIXED in v1.6.1)
« on: September 14, 2013, 07:17:01 am »

Hello,
during my linearization development test I have discovered something strange in the Argyll chartread readings of black patches,
and I'm here to humbly ask for your kind collaboration to confirm if I'm really on to something or if, more simply, I'm missing something.

I have written a post in the color management section you can look at, if you wish,
http://www.luminous-landscape.com/forum/index.php?topic=82133.0
that tell the whole story in more details, step by step (starting from a purplish cast in the icc black point).

I have already sent an e-mail to Graeme Gill, the Argyll author, shortly explaining what I have found, asking for an opinion that I will consider a reference, obviously.

After that I have decided to ask for replication of the issue and for this purpose I'm posting here a simple Argyll test procedure
to verify the Argyll "chartread" capability to read black patches in strip mode.
This test should be really simple to do, but it's best suited for people that already have a basic knowledge and experience using Argyll,
and it will really help me to understand what is happening.


So, let's start.


Argyll RGB printer profile "chartread" test with ColorMunki (but it could be done with other spectro if supported by Argyll)
(PC Win 7 64, Argyll 1.6.0)


My test was done using an Epson R3000 (OEM inks)
and Hahnemuehle Photo Rag Baryta 315 PK (no OBA's high quality cotton photo paper).

I have made a simple test target row of 15 patches including at least two black, one white and some saturated colors.
It will easy fit the long border of a single A4 sheet.

The test target was done using the following Argyll commands:
 targen -v -d2 -G -B2 -e1 -f15 test
 printtarg -v -iCM -h -T360 -p45x297 test

Note that A2 and A15 are the two black patches in the strip "test.tif" I have attached,
so, you don't need to do it.


TEST PROCEDURE:

Download the "Argyll_test.zip" file I have attached.

Print "test.tif" ensuring to disable color management in the application AND in the printer driver.
(I have used ACPU leaving 15mm margin in the left border in the Epson printer settings layout panel)

Wait at least 20 minutes to let ink settle a bit.

Open a command line (I assume Argyll is still installed and ready on your system) and put the test.ti1 and test.ti2 files I have provided
in a folder you will use for this test.
 
Read the row in strip mode with Argyll chartread and rename each time the "test.ti3" for further comparison

 chartread -v -H -L -B -T0.4 test
 rename test.ti3 teststripH.txt

 chartread -v -L -B -T0.4 test
 rename test.ti3 teststrip.txt

Now, read the row (or at least the A2 and A15 patches) in single spot mode and rename the "test.ti3" for further comparison

 chartread -v -H -p -L -B -T0.4 test
 rename test.ti3 testspotH.txt

 chartread -v -p -L -B -T0.4 test
 rename test.ti3 testspot.txt

Open the variuos .txt files above in a spreadsheet or in Notepad++,
look at the Lab values of the two black patches (A2 and A15) and compare it.

You will see that in strip mode the A2 and A15 black are measured with big a* and even bigger b* difference than in spot mode
and is evident that the readings done the -H (high resolution) engaged are the worst in both modes.


Finally, launch X-Rite ColorPicker (provided with ColorMunki) and read the same row in strip mode,
save it as "teststripCM" palette, look on screen at the Lab values of the A2 and A15 patches, write down it
and compare it with the ones already read with Argyll chartread.

Note that, here in ColorPicker, if you additionally read again the same A2 and A15 patches in spot mode (instead of strip mode)
you get nearly the same results (I saved them as "testspotCM"),
and both these readings are similar to the chartread readings made in spot mode (-p option).

Moreover, note that the majority of the colored patches does not appear to be so much different between chartread and ColorPicker (both in strip and spot mode),
so this issue affect mainly the black readings.


Please, try to replicate the test in your setup (with different printer/papers) and, maybe, share your findings.

It would be nice if someone would try to do the same test, by printing the same attached "test.tif" target
but using a i1Pro2 spectro (in UV-cut) instead of the ColorMunki to compare the chartread readings af the A2 A15 black patches
in strip/spot mode, with/without -H option engaged and, maybe, even in their X-Rite Colorpicker equivalent (which I ignore).

I have attached a zip file including my "test.tif" target you can directly use
and all the mentioned Argyll files:
test.ti1 test.ti2 teststripH.txt teststrip.txt testspotH.txt testspot.txt
plus the ColorPicker readings (teststripCM.txt and testspotCM.txt)

In addition, for your convenience, I have attached a pdf file showing all the measurements and highlighting the deviation for the blacks patches with different colors.


I will resume here the blacks reading and my comments:

         L   A   B

chartread strip -H
A2         2.5   -2.1   4.3
A15         3.0   -2.0   5.2

chartread strip
A2         2.7   0.1   3.2
A15         3.0   0.1   3.6



chartread spot -H
A2         3.3   -0.7   1.6
A15         3.7   -1.0   2.7

chartread spot
A2         3.3   -0.2   0.4
A15         3.8   -0.2   0.7



ColorPicker strip
A2         3.7   0.0   -0.5
A15         4.1   -0.1   0.0

ColorPicker spot
A2         3.7   0.1   -0.3
A15         4.2   0.0   -0.1


The bottom line is  that, at least in my setup, there is an evident problem for chartread to correctly read the black patches in strip mode with ColorMunki,
and the so often used -H mode should really be avoided, even in spot mode.
The only acceptable readings from chartread for the blacks are allowed in spot mode (-p) and without -H,
and this is a real pain imho...

Note that it is not an hardware issue of the ColorMunki, because the same strip if read in ColorPicker gives consistent and correct measurement values (both in strip and spot mode).

I have repeated the test several time, using different targets, and I observed always the same results.


Please, let me know your opinion.
Any comment is welcome and will be really appreciated.
Ciao.

Andrea :)









« Last Edit: October 01, 2013, 12:52:03 pm by NeroMetalliko »
Logged

darlingm

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 361
    • Westland Printworks
Re: Argyll chartread test: please try to replicate
« Reply #1 on: September 15, 2013, 04:29:09 am »

I have an i1 Photo Pro 2.  Would it be useful for you for this test to be done on the Epson Stylus Pro 9900 (original Epson inks), on either Breathing Color Vibrance Luster or Gloss photo paper (with OBAs), or B.C. Pura Smooth matte fine art paper (OBA free?)  Might take a few days, photo papers I could do faster.  Matte paper I'd probably wait till I was switching to matte black anyways.
Logged
Mike • Westland Printworks
Fine Art Printing • Amazing Artwork Reproduction • Photography
http://www.westlandprintworks.com • (734) 255-9761

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: Argyll chartread test: please try to replicate
« Reply #2 on: September 16, 2013, 05:55:54 pm »

If this is a major concern it might be better to post it to the ArgyllCMS list serve and see what Graeme and others might have to say on this topic.  I regularly use ArgyllCMS to generate profiles and leave the default # of black patches at 4.  I've never seen anything anomalous in the readings when I look at the values in Excell (and I also include a 51 step B/W patch set as well).  All my readings are done with an i1 Pro.  At one point in the distant past I did a 21 step B/W patch set and read values both using the scan and spot features and didn't see any statistical difference between the readings.  I would have to go back and see what paper I used (probably Museo Silver Rag).

Alan
Logged

NeroMetalliko

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 78
Re: Argyll chartread test: please try to replicate
« Reply #3 on: September 16, 2013, 06:36:32 pm »

Hello Alan,
nice to hear you :)

Graeme Gill (the Argyll author) was kind enough to answer to my mail today,
I have then provided him all the files needed for the test in order to try to replicate the issue.
It could be that the problem arises only with the ColorMunki and not with other spectro,
so probably you don't have to worry about it.
In any case it would be nice to see someone else trying to replicate the test I have prepared and explained above,
just to collect some additional information that could be useful to further investigate/clarify the issue, if any.

Let's wait and see.
Ciao.

Andrea :)
Logged

NeroMetalliko

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 78
Re: Argyll chartread test: please try to replicate
« Reply #4 on: September 17, 2013, 12:37:35 pm »

Hello,
I was asked from Graeme Gill to make another test using a previous version of Argyll and precisely the v1.4.0
because during the change from V1.4.0 to v1.5.0 there was something changed in the chartread and/or usb drivers code
that could be the reason for the issue I have observed.

So I have re-done the test using the previous V1.4.0 revision and I can finally confirm that Graeme Gill was absolutely right. :)

in attachment you can find the zip files containing the ti3 files (renamed to
 txt) of the Argyll chartread simplified test results
 made with V1.4.0 for both strip/spot with/without -H option.

 In addition, for convenience, I have attached a  pdf file resuming the test results,
including the ColorPicker readings which I have still re-done

Looking at the comparison table I made the following considerations:

 - after some days the ink in the printed target settled more (and/or the surface received some little damage)
 so the blacks are a lttle bit less black (near a L* 0,5 shift).
 . overall the matching between X-Rite ColorPicker and Argyll chartread readings is  strongly better in v1.4.0
 - the chartread v1.4.0 produces more reliable and consistent readings between
 their strip/spot mode, but there is still a little deviation, mainly in the a* and b* measured values.
 - the -H option doesn't seem to increase the accuracy at all, maybe the
 opposite, because the readings are still a little bit worse if engaged in both
 strip and spot mode, in particular for the b* values, and this is true even for
 the colored patches if you keep in consideration even the Colorpicker values.
 - The b* readings are the less reliable measurement.
 - The X-Rite ColorPicker is confirmed to be very consistent between their
 strip and spot mode and, from this point of view, it is still slightly better
 than Argyll chartread between their strip and spot mode.
 - In the v1.6.0 chartread .ti3 (renamed to txt) data of the previous test there is definitely something wrong
 (please take a look to the first reflectances showing big negative values for the black patches).

I have already sent all these information to Graeme Gill in order to help him in this task,
let's hope he could fix the issue without too much time expense.

So, now,
the good news is that probably Graeme still knows where the problem could be, and I hope this V1.4.0 vs V1.6.0 comparison could be useful to iron out the issue.

Another good news is that, at least at the moment, we already have an effective workaround:
my opinion is that all the ColorMunki+Argyll users should consider to use (or roll back) the previous Argyll V1.4.0 release instead of the more recent ones
in order to avoid potential problem in the black readings (which produces a "colored" black point in the icc profile).
Note that to successfully roll back to V1.4.0 you have to uninstall the Argyll usb drivers for the Colormunki too.
It is not clear if the issue is confined to the ColorMunki only, probably yes, but I cannot test other spectro because I don't own it :)

That's all, I hope this could be appreciated,
maybe let me know your opinion.

Ciao.

Andrea :)

« Last Edit: September 18, 2013, 03:27:52 am by NeroMetalliko »
Logged

darlingm

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 361
    • Westland Printworks
Re: Argyll chartread test: please try to replicate
« Reply #5 on: September 20, 2013, 05:21:16 pm »

Is it known if the Argyll 1.6 drivers for i1 Photo Pro 2 & equivalent have the same bug?
Logged
Mike • Westland Printworks
Fine Art Printing • Amazing Artwork Reproduction • Photography
http://www.westlandprintworks.com • (734) 255-9761

NeroMetalliko

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 78
Re: Argyll chartread test: please try to replicate
« Reply #6 on: September 20, 2013, 06:20:57 pm »

Is it known if the Argyll 1.6 drivers for i1 Photo Pro 2 & equivalent have the same bug?

Hello,

the only real way to be sure it's trying to replicate the test and see the results,
I cannot do it because I don't own a i1Pro 2.

That said, based on what I have heard from Graeme Gill this should be a specific ColorMunki issue,
so, at the moment, no reason to fear until something strange is not clearly observed/published/replicated.

I can tell you that the issue was already identified and it should be soon officially fixed
restoring the performances of chartread (for ColorMunki in strip mode) to a level similar to the V1.4.0 as per my early test on beta versions.

Ciao.

Andrea :)

« Last Edit: September 20, 2013, 06:23:36 pm by NeroMetalliko »
Logged

NeroMetalliko

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 78
Re: Argyll chartread test: please try to replicate (FIXED in v1.6.1)
« Reply #7 on: October 01, 2013, 12:58:39 pm »

Hello,
after some days of testing different beta versions and the effective work of Graeme Gill
I'm happy to report that the chartread issue I have discovered/illustrated in this thread
was successfully fixed in the new Argyll update v1.6.1 released yesterday,
as visible in the changes log file:


Argyll CMS change log
=====================

Version 1.6.1 (30th September 2013)
-------------
...

*   Fix bug introduced into ColorMunki (spectro) reflective
   measurement that created discrepancies between spot
   and strip measurement mode. Changed short wavelength
   handling to be more like the manufacturers driver in
    duplicating values, reduced reflective strip reading
   scan rate to 55Hz to reduce reflective electrical interference
   error.
...


It was a quite intense work in the past days, but now I can say that I have verified that with this update the performances of chartread in strip mode using ColorMunki are drastically better than in v1.5.0 - v1.6.0 version and slightly better than ever.
The spot/strip mode now is good matching and the overall accuracy/repeatability could be considered on par with the one showed from X-Rite ColorPicker software.

Based on the various test done in these days comparing several chartread releases,
I can add that the -H (high reslolution) mode was not recommended at all with ColorMunki
because the spectrum data was noticeably affected by noise and the overall accuracy/repeatability of the measurements was always a little bit worse
than the one achieved by not engaging the -H flag. This fact was noted in both strip and spot mode (-p flag) but more evidently in strip mode.

NOTE:
currently (v1.6.1) there is still a little bug in chartread because the -H flag is still not correctly working (missing 3nm interval in spectral data output), but it does not seem to harm anymore.
This bug will be fixed in the next v1.6.2 release and I will check if the noise issue will still be there or if Graeme Gill will be able to do its magic again by successfully resolving even this not desired side effect.

In addition I can confirm a slight benefit by using the -T0.4 option:
I have compared readings accuracy/repeatability in strip/spot mode
using -T0.2, -T0.4 and without -T and, even if the difference is small, the best results were obtained by using T0.4 followed by not using -T (-T0,2 was too tight).

Based on all these findings my current (and recommended) command line for chartread print profiles using ColorMunki is:
chartread -v -B -T0.4 'profilename'
 
Note that, as per what is known, this issue was limited only to the ColorMunki spectro.

My humble suggestion is that, if you have made some print profiles using ColorMunki in strip mode with version v1.5.0 to v1.6.0
you should consider to remake the profiles, specially if you have used the -H option.

Some details regarding the specific measurements I have done are available in the attached pdf file resuming a comparison between strip/spot mode of chartread v1.6.1 and ColorPicker, including a 5x times averaging.

I have modified the main title of the thread in order to reflect the 'FIXED' status of the issue.
(Sorry for the duplicated post, but I think it could be useful to give it more visibility to this information).

I Hope this will be helpful.

Ciao.

Andrea :)
Logged
Pages: [1]   Go Up