Severe gamut reduction in profiles Colormunki vs i1Pro2 in ArgyllCMS

[GWGill]

That's quite disturbing. I don't like unexplained phenomina. I've never seen a difference between printing directly from IK1Profiler or the tif file using ACPU. Or, for that matter using the null transform trick to print directly from Photoshop.

The saga of Apple providing a reliable API for applications to print color profile test charts has been going on for something like a decade. It's a complete disgrace, with lots of finger pointing between Apple and printer manufacturers, but seemingly no final resolution. Meanwhile, it all just works on MSWindows ...

[Doug Gray]

To determine where the inconsistencies are it's best to use the same, small chart for both the ColorMunki and I1Pro.

I recommend this 200 Argyll patch set:
targen -v -d2 -G -e0 -B0 -g8 -f200 -R test_target_cm
targen -v -d2 -G -e0 -B0 -g8 -f200 -R test_target
printtarg -v -r -iCM -h -b -t360 -m3 -M3 -P -pA4 test_target_cm
printtarg -v -r -ii1 -h -b -t360 -m3 -M3 -P -pA4 test_target

I've attached the same as a CGATs file that can be imported into I1Profiler as well.

[Doug Gray]

CM and I1Pro2 files added.

[vikcious]

CM and I1Pro2 files added.

Here are the two results from Argyll... with even more reduced gamut! (( I just don't get it!

Paper is RC luster, semi-matt

Commands used:
targen -v -d2 -G -e0 -B0 -g8 -f200 -R test_target_cm
targen -v -d2 -G -e0 -B0 -g8 -f200 -R test_target_i1p2

printtarg -v -r -iCM -h -b -t300 -m3 -M3 -P -pA4 test_target_cm  -> printed with null transform trick
printtarg -v -r -ii1 -h -b -t300 -m10 -M10 -P -pA4 test_target_i1p2 -> printed with null transform trick

chartread -v -AA test_target_cm
chartread -v -AA test_target_i1p2

colprof -v -qh -i D50 -o 1931_2 -r1.0 -cmt -dpp -S "AdobeRGB.icc" -D "ArtMaker Cristal Luster 270g CM" test_target_cm
profcheck -k test_target_cm.ti3 test_target_cm.icm
Profile check complete, errors(CIEDE2000): max. = 0.775805, avg. = 0.286134, RMS = 0.322064

colprof -v -qh -i D50 -o 1931_2 -r1.0 -cmt -dpp -f -S "AdobeRGB.icc" -D "ArtMaker High Gloss 260g i1p2" test_target_i1p2 or
colprof -v -qh -i D50 -o 1931_2 -r1.0 -cmt -dpp -fM2 -S "AdobeRGB.icc" -D "ArtMaker High Gloss 260g i1p2" test_target_i1p2
profcheck -k test_target_i1p2.ti3 test_target_i1p2.icm
Profile check complete, errors(CIEDE2000): max. = 1.777152, avg. = 0.925869, RMS = 0.979589

I am not yet done with the i1Profiler icc creation. I will updated it asap.

[vikcious]

Adding profiles generation logs.

[Doug Gray]

Your latest tests make it possible to compare the patches with the same RGB values. As for the smaller gamut compared to the other 260g profiles, that's from using the tiny patch set and is about what I would expect. But the purpose of the small patch set is to identify differences between the spectrophotometers and workflow.

I'd like to clarify something: Going back to the start, you seem to have two different papers. One is typical of matte paper with a high black point, the other typical of glossy type paper. Is this the case?

"ArtMaker High Glossy 260g" : typical characteristics of glossy paper.
"ArtMaker High Glossy 270g" : typical characteristics of matte paper.

The latest tests are the 270g paper and show typical matte characteristics.

Comparing the two spectro's over the first 8 neutral patches which span RGB 0 to 255, I've attached a spectral plot. The I1Pro shows  more reflectance (lighter levels) in the darker patches. This is consistent with your earlier data too. Could be the I1Pro2 might need to be cleaned. But this isn't a huge effect in comparison to the 260g v 270g profiles and data which indicates either different paper or printer settings.

The two profiles from the I1Pro are the same and characteristic of M0 spectros. You can see the bump up in the shorter wavelengths. To have Argyll simulate M2 you need to give colprof the option "-fM2"

However, uV only affects the shorter wavelengths and has no impact on the reflectance spectrum over about 520nm. So the differences there are due to the spectros themselves. It might be that the I1Pro components have acquired some small amount of contamination. You might try removing the head and cleaning the glass aperture.

The graph shows the log10(reflectance) so -1=10% refl, -2=1% refl, etc.  Otherwise the graph lines would cluster toward the bottom. The green line at the bottom is the very low reflectance typical of glossy papers and if from an earlier 260g profile.

Notice how the I1Pro increasingly diverges from the CM as the patches get darker.

[vikcious]

Hi Doug & others,

Today I have spent an important amount of time trying to re-run all the possible scenarios in a more structured way, easier to follow and reproduce, in order to properly track the results. Since I'm about to lose my mind...

Here we go! 

Printer Model: Canon ImagePrograph PRO-2000 (all original inks)
Paper Type: RC, high OBA, semi-matte (or semi-glossy, whatever!)
Spectrophotometers: ColorMunki Design (aka CM) & ES-2000 / i1 Pro2 (aka i1pro2) -> both clean with i1Diagnostics "ALL PAASED"
Calibration software: i1Profiler v1.8.2 (all modules licensed) & ArgyllCMS v2.0.1

This is the link where all the profiles referenced below can be found:
https://drive.google.com/open?id=1JAZd45TaIKzA5ZWAG_q_Vw2u9fF-LLV7

1iProfiler WORKFLOW
Target Generation
Generated 650 patches target (/w16 gray patches) resulting in TIFF/8bit target

Target Printing
Printed each target on one A4 paper using:
"null trick" version from PS (as guided by you) --> All files with "null_print" in their filename are based on tiff printed with "null trick"
Canon Print Studio Pro from PS /w "No color correction" option (as indicated by Mark Segal's review) --> All files with "psp_print" or "cpsp_print" in their filename are based on tiff printed with Canon's Print Studio Pro

Target Scanning
Dual Scan (M0, M1, M2 and OBC)
Printer Information: RGB Printer
Paper Information: Matte

Profile Generation
Ambient Light: CIE Illuminant D50

Perceptual: Contrast +25
Perceptual: Saturation +25
Perceptual: Neutralize gray +50

Tables: AtoB - Large
Tables: BtoA - Large
Tables: Granularity 16bits

Advanced: Smoothness +50
Advanced: Chromatic Adaptation: Bradford
Advanced: ICC Profile Version: 2 or 4
Advanced: Profile White Point: Default

Profiles created:
i1profiler_650_on_A4_null_print_v2.icm
i1profiler_650_on_A4_null_print_v4.icm
and
i1profiler_650_on_A4_psp_print_v2.icm
i1profiler_650_on_A4_psp_print_v4.icm


ArgyllCMS WORKFLOW
Target Generation
Generated 650 patches target (/w64 gray patches) resulting in TIFF/8bit target

targen.exe -v -d2 -e8 -G -g64 -f650 ArgyllPure_650patches_null-print
targen.exe -v -d2 -e8 -G -g64 -f650 ArgyllPure_650patches_psp-print

printtarg.exe -v -ii1 -R0 -n -t300 -C -M5 -m5 -pA4 -a 0.857 -A 0.857 ArgyllPure_650patches_null-print
printtarg.exe -v -ii1 -R0 -n -t300 -C -M5 -m5 -pA4 -a 0.857 -A 0.857 ArgyllPure_650patches_psp-print

Target Printing
Printed each target on one A4 paper using:
"null trick" version from PS (as guided by you) --> All files with "null_print" in their filename are based on tiff printed with "null trick"
Canon Print Studio Pro from PS /w "No color correction" option (as indicated by Mark Segal's review) --> All files with "psp_print" or "cpsp_print" in their filename are based on tiff printed with Canon's Print Studio Pro

Target Scanning
Slow scanning, using i1pro2 table and guided rule

chartread -v -AA -H ArgyllPure_650patches_null-print
chartread -v -AA -H ArgyllPure_650patches_psp-print

Profile Generation
colprof -v -qh -i D50 -o 1931_2 -r1.0 -cmt -dpp -fM0 -S "AdobeRGB.icc" -D "ArtMaker Cristal Luster NULL ARG_SCAN M0" ArgyllPure_650patches_null-print --> profile: ArgyllPure_650patches_null-print_M0_QH.icm
and
colprof -v -qh -i D50 -o 1931_2 -r1.0 -cmt -dpp -fM0 -S "AdobeRGB.icc" -D "ArtMaker Cristal Luster NULL ARG_SCAN M0" ArgyllPure_650patches_psp-print --> profile: ArgyllPure_650patches_psp-print_M0_QH.icm
colprof -v -qu -i D50 -o 1931_2 -r1.0 -cmt -dpp -fM2 -S "AdobeRGB.icc" -D "ArtMaker Cristal Luster NULL ARG_SCAN M2" ArgyllPure_650patches_psp-print --> profile: ArgyllPure_650patches_psp-print_M2_QU.icm

ADDITIONALLY
1. Saved the measurements from i1Pofiler as i1Profiler CGATS Custom file ".txt"
2. Transformed the files in ArgyllCMS with:
txt2ti3.exe -v -2 argyll_650_on_A4_null_print_M0.txt argyll_650_on_A4_null_print_M0
txt2ti3.exe -v -2 argyll_650_on_A4_null_print_M1.txt argyll_650_on_A4_null_print_M1
txt2ti3.exe -v -2 argyll_650_on_A4_null_print_M2.txt argyll_650_on_A4_null_print_M2
and
txt2ti3.exe -v -2 argyll_650_on_A4_psp_print_M0.txt argyll_650_on_A4_psp_print_M0
txt2ti3.exe -v -2 argyll_650_on_A4_psp_print_M1.txt argyll_650_on_A4_psp_print_M1
txt2ti3.exe -v -2 argyll_650_on_A4_psp_print_M2.txt argyll_650_on_A4_psp_print_M2

3. Created the ICC profiles from the resulting ".ti3" files as:
colprof -v -qh -i D50 -o 1931_2 -r1.0 -cmt -dpp -fM0 -S "AdobeRGB.icc" -D "ArtMaker Cristal Luster 270g ARG_M0_2_M0" argyll_650_on_A4_null_print_M0  --> profile: argyll_650_on_A4_null_print_M0.icm
colprof -v -qh -i D50 -o 1931_2 -r1.0 -cmt -dpp -fM0 -S "AdobeRGB.icc" -D "ArtMaker Cristal Luster 270g ARG_M1_2_M1" argyll_650_on_A4_null_print_M1  --> profile: argyll_650_on_A4_null_print_M1.icm
colprof -v -qh -i D50 -o 1931_2 -r1.0 -cmt -dpp -fM0 -S "AdobeRGB.icc" -D "ArtMaker Cristal Luster 270g ARG_M2_2_M2" argyll_650_on_A4_null_print_M2  --> profile: argyll_650_on_A4_null_print_M2.icm
and
colprof -v -qh -i D50 -o 1931_2 -r1.0 -cmt -dpp -fM0 -S "AdobeRGB.icc" -D "ArtMaker CristalLuster NULL ARG_M0_2_M0" argyll_650_on_A4_null_print_M0  --> profile: argyll_650_on_A4_psp_print_M0.icm
colprof -v -qh -i D50 -o 1931_2 -r1.0 -cmt -dpp -fM0 -S "AdobeRGB.icc" -D "ArtMaker CristalLuster NULL ARG_M1_2_M1" argyll_650_on_A4_null_print_M1  --> profile: argyll_650_on_A4_psp_print_M1.icm
colprof -v -qh -i D50 -o 1931_2 -r1.0 -cmt -dpp -fM0 -S "AdobeRGB.icc" -D "ArtMaker CristalLuster NULL ARG_M2_2_M2" argyll_650_on_A4_null_print_M2  --> profile: argyll_650_on_A4_psp_print_M2.icm

CONCLUSIONS
After all this effort, time spent, resources wasted I still cannot get a better profile from the i1 Pro2 as compared to ColorMunki.
Two profiles tell the whole story:
ArtMakerCristaLuster_840p_HQ.icm --> 658K cubit units... better DMax
vs
ArgyllPure_650patches_psp-print_M2_QU.icm --> 532K cubic units (a decrease of 20% in gamut size!!!)

Aside from these dry technicalities the overall quality in image conversion (on screen) and image printed (visual arbitration) is obvious, not neglect-able and bothering!

  What is going on?!! Please help!  

[Doug Gray]

Check the google drive
The only thing that shows up at this link is the tif target file.

https://drive.google.com/open?id=1MQ-YyKfVDGh9tPoRw3wBp2PfXrRIfL_m

Excellent set of data. Look forward to seeing all those files and profiles. Should be possible to figure out what's going on.

BTW, the null=transform trick is something I use only for printing targets with Photoshop that can't be created/printed in I1Profiler. And I always check against new versions of PS given their warning.

[vikcious]

Check the google drive
The only thing that shows up at this link is the tif target file.

https://drive.google.com/open?id=1MQ-YyKfVDGh9tPoRw3wBp2PfXrRIfL_m

Excellent set of data. Look forward to seeing all those files and profiles. Should be possible to figure out what's going on.

BTW, the null=transform trick is something I use only for printing targets with Photoshop that can't be created/printed in I1Profiler. And I always check against new versions of PS given their warning.

Oh dear, my bad! I have updated the link!

[vikcious]

BTW, the null=transform trick is something I use only for printing targets with Photoshop that can't be created/printed in I1Profiler. And I always check against new versions of PS given their warning.

Hmm... looking at the results of test image ("PrinterEvaluationImage_V002_ProPhoto.tif") converted to the created profiles, I can clearly see that the "null trick" printing is not the way to go! It badly affects smooth transitions and causes a lot of black blocking. I could come back with updated shots to exemplify.

[Doug Gray]

Hmm... looking at the results of test image ("PrinterEvaluationImage_V002_ProPhoto.tif") converted to the created profiles, I can clearly see that the "null trick" printing is not the way to go! It badly affects smooth transitions and causes a lot of black blocking. I could come back with updated shots to exemplify.

If you are using the null transform trick it's important not to use the Canon plug-in. Use only the driver and manually disable color management. You would use the same settings as printing directly from I1Profiler or ACPU.

[vikcious]

If you are using the null transform trick it's important not to use the Canon plug-in. Use only the driver and manually disable color management. You would use the same settings as printing directly from I1Profiler or ACPU.

That is exactly what I did! Prior to clicking "Print" I would always check to see if color correction is off in the driver settings.

[Doug Gray]

That is exactly what I did! Prior to clicking "Print" I would always check to see if color correction is off in the driver settings.

Looks right, assuming you had assigned the same "Adhesive" profile to the target.

For printing a regular image say, in Adobe RGB, you would first convert from Adobe RGB into the desired printer profile setting intents as desired then select that same profile to print it from Photoshop similar to your example.

You should see exactly the same results printing with ACPU (Which is the gold standard for printing w/o CM).

Not sure what's going on. Perhaps the Canon plug-in is somehow altering the printing. While I have a Canon, an older 9500 II, I haven't been able to get their plug-in to work in current Photoshops so can't test that.

[Doug Gray]

Here's a summary of white/black points for the new set of profiles:

The 3 numbers are L*a*b* values in 3 groups. "WP" is the Lab value of unprinted paper. The b* value will be quite negative for papers with high OBA for M0 and especially M2 profiles because of the uV.

BP_PI is the Lab for printed RGB (0,0,0) using Perceptual Intent. It's usually more neutral than the black using Relative Intent.

BP_RI is the printed RGB (0,0,0) without BPC. It's usually the lowest possible L* but may have more chromaticity than Perceptual intent.

The "Descr" is the name of the profile as shown in applications and is not necessarily the same as the profile file name.


Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\ArgyllPure_650patches_null-print_M0_QH.icm"
Descr: "ArtMaker Cristal Luster NULL ARG_SCAN M0" Ver: 2.2.0   Copyright: "Copyright, the creator of this profile"
WP:  92.0   2.5  -9.7     BP_PI:  17.0   0.8   2.1    BP_RI:  16.5   1.3   0.9

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\ArgyllPure_650patches_psp-print_M0_QH.icm"
Descr: "ArtMaker Cristal Luster PSP ARG_SCAN M0" Ver: 2.2.0   Copyright: "Copyright, the creator of this profile"
WP:  91.9   2.6 -10.0     BP_PI:  14.2   0.3  -0.2    BP_RI:  14.3   0.4  -0.4

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\ArgyllPure_650patches_psp-print_M2_QU.icm"
Descr: "ArtMaker Cristal Luster PSP ARG_SCAN M2" Ver: 2.2.0   Copyright: "Copyright, the creator of this profile"
WP:  91.7   0.7  -4.2     BP_PI:  14.2  -0.5   2.7    BP_RI:  14.2  -0.3   1.8

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\ArtMakerCristaLuster_840p_HQ.icm"
Descr: "ArtMaker Cristal Luster SemiGloss" Ver: 2.2.0   Copyright: "Copyright, the creator of this profile"
WP:  92.7   1.2  -4.2     BP_PI:  11.7   0.6   1.7    BP_RI:  11.7   0.6   1.7

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\argyll_650_on_A4_null_print_M0.icm"
Descr: "ArtMaker CristalLuster NULL ARG_M0_2_M0" Ver: 2.2.0   Copyright: "Copyright, the creator of this profile"
WP:  92.0   2.5  -9.5     BP_PI:  17.3   1.1   1.7    BP_RI:  17.0   2.1  -0.8


Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\argyll_650_on_A4_null_print_M1.icm"
Descr: "ArtMaker CristalLuster NULL ARG_M1_2_M1" Ver: 2.2.0   Copyright: "Copyright, the creator of this profile"
WP:  92.4   3.5 -14.1     BP_PI:  17.4   1.2   0.6    BP_RI:  17.0   2.8  -3.1

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\argyll_650_on_A4_null_print_M2.icm"
Descr: "ArtMaker CristalLuster NULL ARG_M2_2_M2" Ver: 2.2.0   Copyright: "Copyright, the creator of this profile"
WP:  91.7   0.6  -2.2     BP_PI:  17.2   0.7   2.7    BP_RI:  16.9   1.6   0.8

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\argyll_650_on_A4_psp_print_M0.icm"
Descr: "ArtMaker CristalLuster PSP ARG_M0_2_M0" Ver: 2.2.0   Copyright: "Copyright, the creator of this profile"
WP:  91.4   2.6  -9.5     BP_PI:  14.1   0.6  -0.3    BP_RI:  14.1   0.6  -0.3

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\argyll_650_on_A4_psp_print_M1.icm"
Descr: "ArtMaker CristalLuster PSP ARG_M1_2_M1" Ver: 2.2.0   Copyright: "Copyright, the creator of this profile"
WP:  91.8   3.5 -14.1     BP_PI:  14.2   0.8  -1.4    BP_RI:  14.2   0.8  -1.4

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\argyll_650_on_A4_psp_print_M2.icm"
Descr: "ArtMaker CristalLuster PSP ARG_M2_2_M2" Ver: 2.2.0   Copyright: "Copyright, the creator of this profile"
WP:  91.1   0.7  -2.3     BP_PI:  14.0   0.4   0.5    BP_RI:  14.0   0.4   0.5


Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\i1profiler_650_on_A4_null_print_v2.icm"
Descr: "650_on_A4_null_print_v2.icm" Ver: 2.1.0   Copyright: "Copyright X-Rite, Inc."
WP:  91.9   2.6  -9.6     BP_PI:  16.8   2.8  -2.1    BP_RI:  16.8   2.8  -2.1

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\i1profiler_650_on_A4_null_print_v4.icm"
Descr: "650_on_A4_null_print.icm" Ver: 4.3.0   Copyright: "Copyright X-Rite, Inc."
WP:  91.9   2.6  -9.7     BP_PI:  16.8   2.8  -2.1    BP_RI:  16.8   2.8  -2.1

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\i1profiler_650_on_A4_psp_print_v2.icm"
Descr: "650_on_A4_psp_print_v2.icm" Ver: 2.1.0   Copyright: "Copyright X-Rite, Inc."
WP:  91.3   2.7  -9.6     BP_PI:  14.0   0.3  -0.4    BP_RI:  14.0   0.3  -0.4

Profile: "C:\WINDOWS\System32\Spool\Drivers\Color\i1profiler_650_on_A4_psp_print_v4.icm"
Descr: "650_on_A4_psp_print_v4.icm" Ver: 4.3.0   Copyright: "Copyright X-Rite, Inc."
WP:  91.3   2.7  -9.7     BP_PI:  14.0   0.3  -0.4    BP_RI:  14.0   0.3  -0.4


One thing that's clear. There is a difference between printing bypassing the plug-in. And the plug-in appears to print darker blacks.

Have you tried printing directly from I1Profiler rather than making the tif then printing it in Photoshop with PSP? It may be that PSP is  needed for best results. Perhaps the driver is different which may well be the case since it processes 16 bit data through the PSP. If so you should always print through PSP.

Aside from that, from earlier results it appears the I1Pro 2 is also higher reading L*  than the CM. You may need to make profiles with the CM or spring for a new I1Pro2.

I've attached an image of the gamut slice at L*=27 for the I1Profiler PSPS v Null. The null trick profile (white outline) is extremely ratty.

I've never seen anything quite like that but the only Canon I have is the 9500II and for some reason PSP does not work with it and Windows at all. Printing directly to the driver from I1Profiler as well as the null trick works perfectly on it as well as on my Epson 9800. This strongly suggests there are different drivers in operation using PSP and printing directly though the driver which is limited to 8 bits by Windows.

Added: I just reviewed Mark Segal's review of the 2000 and he notes that ACPU does a pretty bad job on the 2000 (Unlike other printers) and that Canon confirmed and recommended PSP. ACPU, I1Profiler's direct printing, and the null-trick should all use the same path through the printer. So it does appear there is a difference in drivers and PSP should be preferred.

I've found the null trick most useful in making collages where some of the images were converted using Perceptual while others were Rel. Col. or even Abs (which is useful for plopping in things like an accurate CC image). That can still be done but after converting to printer space the whole thing should be untagged and printed with PSP w/o color management.

[vikcious]

You should see exactly the same results printing with ACPU (Which is the gold standard for printing w/o CM).

You don't wanna know how poor the targets printed with ACPU really looked! All the gray scale patches are contaminated with some kind of brownish tint... I was even afraid of scanning it!
On a top 3 of visually evaluated targets 1) Canon PSP 2) Null trick printing 3) ACPU 

[vikcious]

Dear Doug, yet again, thanks a lot for the effort you've put into shedding some light into this weird story.

Here's a summary of white/black points for the new set of profiles:

The 3 numbers are L*a*b* values in 3 groups. "WP" is the Lab value of unprinted paper. The b* value will be quite negative for papers with high OBA for M0 and especially M2 profiles because of the uV.
BP_PI is the Lab for printed RGB (0,0,0) using Perceptual Intent. It's usually more neutral than the black using Relative Intent.
BP_RI is the printed RGB (0,0,0) without BPC. It's usually the lowest possible L* but may have more chromaticity than Perceptual intent.
The "Descr" is the name of the profile as shown in applications and is not necessarily the same as the profile file name.

Could you please share how / where from (what tool?) do you use to extract all this info? I would like to be able to do it myself and not have to bother you every time! Thanks!

One thing that's clear. There is a difference between printing bypassing the plug-in. And the plug-in appears to print darker blacks.

To me it's 100% clear! At least for the PRO-2000+ series the only valid way to print targets remains through the Canon PSP plugin from PS. Even my "best" profile so far (CM based one) used this approach for the targets printing.

Have you tried printing directly from I1Profiler rather than making the tif then printing it in Photoshop with PSP? It may be that PSP is  needed for best results. Perhaps the driver is different which may well be the case since it processes 16 bit data through the PSP. If so you should always print through PSP.

Despite serious doubts, I will probably try.

Added: I just reviewed Mark Segal's review of the 2000 and he notes that ACPU does a pretty bad job on the 2000 (Unlike other printers) and that Canon confirmed and recommended PSP. ACPU, I1Profiler's direct printing, and the null-trick should all use the same path through the printer. So it does appear there is a difference in drivers and PSP should be preferred.

Aside from that, from earlier results it appears the I1Pro 2 is also higher reading L*  than the CM. You may need to make profiles with the CM or spring for a new I1Pro2.

This hurts! Let's discuss this situation a bit.
The EFI ES-2000 was purchased as new, from ebay, (really only showed like 3 seconds of lamp usage at first usage of i1Diagonstics) and I can hardly believe it's a faulty one. Or maybe I'm just reprieving myself this thought! Could it really be?

A) So basically my old CM is able to both read a higher WP and a lower BP against the i1Pro2, correct?

B) Am I correct to say that based on the metrics you've provided i1Pro2 spectrophotometer looks like it's not coping well with dark tones / blacks?

C) Is the most of the clipping, I am seeing from the created profiles, appear to come in dark violet / dark blues values which happen to fit with the way the gamut it's actually shaped and where the big missing chunk appears, when compared against the Canon's or CM's profiles?

D) I am not exactly sure how much the i1Pro2 metering is deviating since I don't have another one at hand, nor do I have anyone close to me with whom I could double-check using same paper.

E) Is there any more conclusive battery of tests I could run for i1Pro2 in order to prove it's really a faulty copy? I kinda feel I need to check more but I am totally clueless on how/what could I do. Any idea?

F) What should I be doing now? Sell it? Try to send it for factory calibration though I am not sure if I should send it to EFI or X-Rite and I suspect it will cost a small fortune to have it calibrated once again, am I wrong?

Looking forward for your (and others) guidance on this issue!

[Doug Gray]

Dear Doug, yet again, thanks a lot for the effort you've put into shedding some light into this weird story.

I was originally taken aback by the differences between your glossy results from "ArtMakerHighGlossy260" and the later matte like paper as the glossy had pretty low L* (3 ish) while the matte was in the teens. It's now clear that this was just different papers somehow as both the I1Pro2 and CM read reasonably low though the CM was a bit lower.

Could you please share how / where from (what tool?) do you use to extract all this info? I would like to be able to do it myself and not have to bother you every time! Thanks!

The data I printed out was from a Matlab function I wrote. I use it to quickly check black points on my own profiles but a broader use is to check the performance of various paper/printer combos by downloading their ICC profiles and running them through the function. My use is like this:

ShowProfileStats('Argyll')

It goes through my installed profiles and anything with 'Argyll' in the profile description or file name gets analyzed and printed. I also use it to identify profiles that are incorrectly generated. For instance Epson's canned profiles typically bake in BPC in the Relative Colorimetric tables which has certain negative effects. Especially printing Abs. Col. when I need to print a precise color.

To me it's 100% clear! At least for the PRO-2000+ series the only valid way to print targets remains through the Canon PSP plugin from PS. Even my "best" profile so far (CM based one) used this approach for the targets printing.
Despite serious doubts, I will probably try.

Not worth it given ACPU gives such bad results. I've looked at the Canon plug-in profiles and it's pretty clear that the plug-in overrides the native print driver. Especially as it also has features to reverse some of the low pass effects that are somewhat intrinsic in print drivers.

This hurts! Let's discuss this situation a bit.
The EFI ES-2000 was purchased as new, from ebay, (really only showed like 3 seconds of lamp usage at first usage of i1Diagonstics) and I can hardly believe it's a faulty one. Or maybe I'm just reprieving myself this thought! Could it really be?

You might be OK. While the differences seem a bit more than I would expect all spectrophotometers differ somewhat. I'm going to compare the CM and I1P profiles, both using PSP, and seen how they compare with the data I have (an I1P2 and iSis XL 2). I recall there was some differences there but it's been a while so I'm not sure how they compare to your differences. The iSis is similar to the CM in that they both natively use a white LED to create M2 files.

A) So basically my old CM is able to both read a higher WP and a lower BP against the i1Pro2, correct?

B) Am I correct to say that based on the metrics you've provided i1Pro2 spectrophotometer looks like it's not coping well with dark tones / blacks?

C) Is the most of the clipping, I am seeing from the created profiles, appear to come in dark violet / dark blues values which happen to fit with the way the gamut it's actually shaped and where the big missing chunk appears, when compared against the Canon's or CM's profiles?

D) I am not exactly sure how much the i1Pro2 metering is deviating since I don't have another one at hand, nor do I have anyone close to me with whom I could double-check using same paper.

E) Is there any more conclusive battery of tests I could run for i1Pro2 in order to prove it's really a faulty copy? I kinda feel I need to check more but I am totally clueless on how/what could I do. Any idea?

F) What should I be doing now? Sell it? Try to send it for factory calibration though I am not sure if I should send it to EFI or X-Rite and I suspect it will cost a small fortune to have it calibrated once again, am I wrong?

Looking forward for your (and others) guidance on this issue!

If you're game, I can generate and post test charts. If you print them, let them dry a day or two, I can scan them with the iSis (XRite's high end spectro) and we can find out how close your I1P and CM profiles are to the iSis. I can PM you my mailing address. It might take some time for the mail but it would be interesting. As an aside, I can make a large patch set of profiles from them to make it worth your effort. From my POV the printer you have appears quite good and I may wind up buying the 1000 for my office. The data would help me evaluate the possible upgrade from my 9500 II which is getting long in the tooth. The 9800 has proven a durable and stable beast so I'm unlikely to upgrade it any time soon.

[Doug Gray]

Well, this is interesting. The ColorMunki profile based on 200 patches is a much closer match to the 650 patch null profile than it is to the PSP.

Algorithm: Use 10,000 random RGB values and convert to Lab values using three profiles. Then make a histogram of the distribution of delta E's between the profiles. This is much more useful than  gamut volume.

CM (200 patch): ArtMaker Cristal Luster 270g CM
PSP I1Pro: ArtMaker Cristal Luster PSP ARG_SCAN M2
Null I1Pro: ArtMaker CristalLuster NULL ARG_M2_2_M2

The CM and Null I1Pro are far closer to each other than the two I1Pro profiles. This suggests that the CM target was printed the same way as the I1Pro Null target.

So in this case the largest differences by far were PSP/Null related and not spectro differences. And the CM profile even had far fewer patches!

Something strange is going on and you might be best to focus on the printing process. Make sure you are using identical settings in PSP and that you don't do a printer calibration in between.

I suggest the following things to do apples to apples.

1. Only generate M2 profiles. Comparing CM and I1Pro is only possible with M2.
2. Don't bother generating separate profiles with I1Profiler and Argyll. They produce effectively the same Relative Colorimetric tables.
3. When using PSP, only print using PSP with identical settings.

If you want to explore Null trick, ACPU, or direct printing from I1Profiler, use the same settings by creating a name for the all the settings then make sure you select that name each time you print. It's possible, perhaps likely, that PSP is changing some of the values in the driver beyond just disabling color management.

[vikcious]

Well, this is interesting. The ColorMunki profile based on 200 patches is a much closer match to the 650 patch null profile than it is to the PSP.

Algorithm: Use 10,000 random RGB values and convert to Lab values using three profiles. Then make a histogram of the distribution of delta E's between the profiles. This is much more useful than  gamut volume.

CM (200 patch): ArtMaker Cristal Luster 270g CM
PSP I1Pro: ArtMaker Cristal Luster PSP ARG_SCAN M2
Null I1Pro: ArtMaker CristalLuster NULL ARG_M2_2_M2

Again Doug working his math magic!

The CM and Null I1Pro are far closer to each other than the two I1Pro profiles. This suggests that the CM target was printed the same way as the I1Pro Null target.

How I hate to disappoint you... Prior to this topic I have never used either "null trick" or ACPU! 110% sure that the target for the CM was printed in PSP! The only difference I can think of was that the targets are tiff/16bits (see below for archive reference)
 

So in this case the largest differences by far were PSP/Null related and not spectro differences. And the CM profile even had far fewer patches!

Well... disappointment doubles: CM used 840 patches... I have uploaded the "ArtMaker Cristal Luster 270g.zip" archive with all the Argyll generated files!

Something strange is going on and you might be best to focus on the printing process. Make sure you are using identical settings in PSP and that you don't do a printer calibration in between.

I'm digging myself a grave to put myself for some rest!

I suggest the following things to do apples to apples.
1. Only generate M2 profiles. Comparing CM and I1Pro is only possible with M2.
2. Don't bother generating separate profiles with I1Profiler and Argyll. They produce effectively the same Relative Colorimetric tables.
3. When using PSP, only print using PSP with identical settings.

Not only am I going to do so but I will screen record the steps to avoid any processing doubts, ok?

If you want to explore Null trick, ACPU, or direct printing from I1Profiler, use the same settings by creating a name for the all the settings then make sure you select that name each time you print. It's possible, perhaps likely, that PSP is changing some of the values in the driver beyond just disabling color management.

Btw... there is only one thing that I remember changing ever since I started i1pro2 instead of CM: I have given up using the "preconditioning" step in the targen syntax! Could that be that it induces any change?
I used to have it like:
targen.exe -v -d2 -c CN_PRO-2000_520_PhotoPaperProLuster.icc -e8 -G -g64 -f840 ArtMakerCristaLuster

I will try to use the same preconditioning with the new i1pro2 targets to see if it brings any changes.

[Doug Gray]

Well... disappointment doubles: CM used 840 patches...

The CM profile I used for that comparison was quite close to the I1Pro2 profile was:"test_target_cm.icm", which was made with 200 patches and has the embedded description: "ArtMaker Cristal Luster 270g CM". The I1Pro profile it was close to has the file name:"argyll_650_on_A4_null_print_M2.icm" and description:"ArtMaker CristalLuster NULL ARG_M2_2_M2"
[/quote]

Don't change too many things at once. It's now clear from the other thread you started that the basic problem is variation in printer settings using PSP.

Your I1Pro 2 is probably behaving perfectly well.