i1 Studio vs. Datacolor Spyder Print Profile comparisons

I got into contact with a member of the German Druckerchannel forum discussing this and that, profiling, the effect of the gloss optimizer, 3rd party vs. genuine Epson inks and the accuracy of profiling software packages in comparison. Let me call this member 'Robert', we continued our discussions by email and exchanged prints, profiles and such by mail for mutual inspection. Robert is using the Spyder Print V.5 package to create icc-profiles and was very much interested to understand how accurate these profiles are. There are basically no tools (freeware) to do such an evaluation. I'm using an old 3D-tool - Monaco GamutWorks - to display gamut properties , and to a smaller degree the old Gamutvision software. Both are old and only support V.2 icc-profiles, I'm not aware of any newer affordable inspection software for V.4 profiles. Anyway - this was a starting point, Datacolor is not offering much in this respect to their users. Robert is not really happy with the stripe reading mode of the SpyderPrint software and the quite frequent data/scan rejects by the software which requires him to reread a particular stripe. I'm using the X-Rite i1Studio package with the i1io robot arm to scan the target sheets which a significant advantage - this hardware offers a fixed scan speed with a very low level of rejects. Most rejects are of the type - ' colors deviate significantly from the target values - do you want to continue ?'. This may happen on normal paper with a small gamut - smaller than the software is apparently expecting. Some other error conditions trigger an automatic rescan at a slower speed for a particular stripe.
I'm not going into any technical details and differences between the Datacolor colorimeter and the X-Rite Si1 Pro or i1Pro2 spectrophotometer.
I did a simple test with Robert - we wanted to profile regular normal copy paper for comparison - I was using a P400 for this purpose . I printed the X-Rite target sheet and as well the Datacolor target sheet from a .tiff file Robert gave me. I sent the print out to him to let Robert process the profile with his software, and he returned the icc-profile to me he created with his Spyder software. We were using the same paper with the same inks on the same printer this way.
These are some of the findings:



This is the gamut of normal paper printed on a P400 and profiled with the i1 software, the volume is pretty small but there are otherwise no particular problem areas, the shape is typical, the surfaces and edges are smooth and even. This is just a 2D view, but there are no problems hidden on the back side of the gamut volume either.


This is the gamut of the icc-profile as created with the Spyder package - and what a surprise - there is a big hole in the gamut volume which I never have seen before like that. The red border outline at the cut through the gamut at a luminance of about L*=45 shows a significant indentation. This is a pretty significant problem with this profile although the Spyder software did not report any unusual problem during the measurement and calculation process. And there is some other deformation visible in the upper part of the gamut.
Please let me continue with another posting to show you another picture of this profile.

A distortion of a profile that serious makes it unusable, the question is how a user can identify such a defunct profile.
I did a display of this profile as well in Gamutvision, the overall look is a kind of similar, the projection of several lightness curves onto the ground show as well these serious indentations , and as well the deformation in the upper part of the gamut.



I discussed this with Robert, there weren't any unusual circumstances as he was scanning the target sheet, he had to do a few re-scans as requested by the software, but otherwise typical as always. The software was accepting the data and calculated the profiles as show above.
I asked him to re-scan the target sheet for comparison to see how and whether handling could cause such variances.
This is the gamut as generated from this 2nd scan, the hole is gone but other serious problems become visible this time



It's not the bottom part affected this time but the upper part - just look to the irregular green line in the right plot which is supposed to be the gamut limit at a luminance of about L=70 - it's nonsense.



Just look to the bottom of the 3D display here with Gamutvision - the projected gamut limits look pretty similar - very much irregular making this profile useless as well.

I found a test report published on the internet addressing these issues
https://www.imaging-resource.com/ACCS/S3P/S3P.HTM

This was a test of the SpyderPrint version 3, Robert is using V.5 at this time, but the problems are apparently not really gone since then. It is some kind of handling issue with the Colorimeter along the ruler - keeping constant speed whatever. Robert did another target scan - by spot reading which takes quite some more time - that finally resulted in a profile which was almost identical with the one I created with the X-Rite i1 software.
We checked a few other profiles Robert created earlier at a time he was not aware of these target scanning problems in stripe mode, he is using a P800 and is very much interested in serious B/W printing.
Here is another profile which we found with some deviations - not as serious as the example above but not really clean at the top



Or here another one



Here is the yellow-green corner affected.

Stay tuned - there is more to come - I need to pull some data together.

Are these issues with the Spyder due at all to the calorimeter in need of a re-calibration, if that can be done at all?

Like spectrophotometers, a colorimeter can experience significant drift dependent upon age and environment independent of actual use.

The described effects are not calibration issues, there is a calibration tile embedded in the unit holder, and you do a cal before you do a scan, very similar to what the i1 software is doing as well. Such erratic readings are not related to calibration but one way or the other to the handling when you are in stripe mode - uniformity of speed, pressure whatever else can cause them - I don't know and Robert does not know either. The single patch mode takes more time and needs a click for every patch , and the profile is identical with the i1 profile in this case. The problem is that the user does not have a good means to see, to be alerted when such reading errors occur. The colors of the patches are displayed as the scanning proceeds - the colors as scanned are different to the theoretical colors which is typical, but you cannot tell which data are outliers or usable for the profile generation.

After we sorted out the unreliable Spyder profile generation - there is a high risk of erratic data with the stripe reading mode - we went on to compare prints , inks, gloss and all that. Robert is preferring B/W prints - either on a Hahnemühle Lustre or a Photo Rag paper. He was using Lyson HD refill inks on his P800 at that time but was not really happy with them - some remaining bronzing on the Lustre paper in the dark/shadow areas. I did some gloss opt. overprints for him to compare - the GO takes away the bronzing but changes the look of the unprinted white areas on the Lustre paper which he did not like so much. He switched back to genuine Epson Ultrachrome HD inks which exhibit visibly less bronzing in the shadow areas w/o GO. This part of the testing was not related to profiles - Datacolor or i1, but since Robert was scrutinizing as well the shadow areas on prints done with different profiles - done with SpyderPrint or i1 which I did for him, we effectively were doing remote profiling each others printers - I did some i1 profiles for him, and he did some Spyder profiles for my P400 as well. We compared the rendering of details in the darker/shadow areas and judged the overall impression - tonal neutrality etc and came to the conclusion that the Epson P800 does a visibly better job than my P400 for B/W printing - which was not so much a surprise but more a confirmation of a previous assumption. The P400 does not use light black inks and does not have any extended B/W printing support via the driver like the P800 does.
Since B/W printing was the main focus of testing we tried to use as well some tools to get a better view what is happening in the shadow areas. It's not easy, there is only one tool providing at least some information - Gamutvision with the display option 'Output vs. Input L*a*b*c*' for different rendering intents and with black point compensation BPC on/off.
There is quite some stuff published on the internet about rendering intents and BPC
http://www.gamutvision.com/docs/blackpoint.html
or more theoretical http://www.color.org/WP40-Black_Point_Compensation_2010-07-27.pdf or by Adobe
http://www.color.org/AdobeBPC.pdf
These documents reflect the industry wide understanding of these parameters .
The profiling of a copy paper reveiled some serious flaws in the Spyder profiling flow so I took a profile for normal paper again and displayed the output vs. input luminance response



Just follow the lower curve - the output luminance equals the input luminance until it reaches the black level of the paper, and the output levels off at this level. The black level of this normal paper is poor with L*~35 (on the output axis) so every color darker than that will disappear in the dark. The black level compensation was introduced longer time ago to compensate this effect and to smooth out the response curve in the shadow range - the upper curve with BPC on. This is the typical behaviour of icc-profiles, I looked up a range of those created with different software, Hahnemühle is publishing profiles for their papers apparently created some time ago with the ProfileMaker package, the predecessor of the i1 Studio software by X-Rite. @mikling is using the well known ArgyllCMS software for the profiles he is publishing on the precisioncolor website. I found some other profiles made with MonacoProfiler software, an even earlier version by GretagMacbeth before they were taken over. farbenwerk is supplying profiles created with the i1 package.
But when displaying such graph with a Spyder profile you'll detect that Datacolor is not following the standard approach

The lower curve shows the B/W response without BPC activated, it already looks like being coded with BPC on, and if BPC gets turned on in the Gamutvision display the response curve jumps up dramatically - from a black point of about L*~35 to 55 - what is going on ? The same effect can be observed with other Spyder profiles - the black point jumps up with BPC on, and the diagram w/o BPC does not compare to the curves of i1 profiles.

It took us quite a long time to understand this, the Spyder support line was not very helpful in this case, and I finally found a Datacolor document - How to print from Adobe Photoshop with Spyder profiles
http://www.datacolor.eu/fileadmin/datacolor/Documents/manuals/Drucken-mit-ICC-in-PS.pdf
( I couldn't find the corresponding document in English)

There is a remark on page 3

Renderpriorität: Bei unseren, mit SpyderPRINT erstellten Profilen, empfehlen wir “Sättigung”, bei Profilen von anderen Herstellern verwendet man entweder „Perzeptiv“ oder „relativ Farbmetrisch“. Tiefenkompensierung: deaktivieren (Häkchen wegklicken)

Use the rendering priority 'Saturation' and deactivate BPC which is quite an unusual setting when you are used to use either perceptual or rel. colorimetric with BPC on.

I could not find any other or more explanation why Datacolor is doing it this way, a Datacolor profile does not provide you with the option to use the profile uncompensated at all.



Here again for comparison
the lower line shows the i1 profile - colorimetric - BPC off
the middle line shows the Spyder profile - colorimetric - BPC off
the upper line shows the Spyer profile - colorimetric - BPC on - which should not be used as per Datacolor instructions.
Datacolor implements some kind of black point compensation into their profile as standard data which you cannot turn off, and Datacolor uses the saturation rendering intent as the default mode for their profiles.
And there is evidence for this unusual setting, an icc-profile consists of a set of tables 2x3 tables for the color transformation for the different rendering intents, and the profile header provides some information for the print software where those tables can be found. You can display these profile properties



and you can see that the table entry points- offsets - for the A2B0 and A2B2 tables - perceptual and saturation - are the same - Datacolor does not differentiate between the perceptual and saturation rending intent.

Let me conclude this evaluation here with the strong advise -
use the saturation rendering intent w/o BPC for printing with Datacolor profiles - all other settings will screw up your print output.

Ink stained Fingers said:

How to print from Adobe Photoshop with Spyder profiles

Click to expand...

I found the English version of the quoted document here

https://www.datacolor.com/wp-content/uploads/2017/02/Print-with-ICC-in-PS.pdf

You'll find the advise on page 3 - to use the saturation rendering intent with BPC off. There is no further explanation why Datacolor deviates here form the industry typical approach other companies are following. This makes it tricky and error prone when you use profiles of different origins.

Another document describes the overall process for profile generation
https://digitalblackandwhite.co.uk/datacolor-spyderprint.html but this seems to cover a previous version and does not address any specific problems with the scanning process.

A magnificent write up of a well researched difficult topic. Well done, ISF.

Please let me report what else we did to understand these profile differences; Robert is using a Hahnemühle Photo Rag for his B/W prints, I profiled this paper for him for his P800 with the i1 software, and he created the corrresponding Spyder profile - without serious kinks. The German Standards Institute DIN has a department for the standardization of colors for the printing , textile and chemical/paints industry sections, and there is a RAL set of such standardized colors - a kind of similar to the Pantone color sets. You can get these colors references in various formats - printed on fans or in a spiral bound booklet or as individual color reference cards larger or smaller - semimatte or glossy . And the RGB and Lab values for these colors are available as well. Robert has access to those reference cards and created a file with some - 6 of these colors - various gray colors , not all of them are neutral like navy gray or concrete gray . He printed these patches with the P800 with the i1 profile on this PhotoRag paper - BPC off and on and with the Spyder profile and compared the prints in a norm light box with the RAL color cards.

He got basically these results :

- the i1 profile gave the best congruence of the color tints and the lightness of the patches - w/o BPC
- the i1 profile with BPC on was printing the patches just slightly lighter - barely visible - but he could expect that from the B/W curve in Gamutvision
- the Spyder profile was printing generally lighter but with otherwise correct colors

Have a look to the Gamutvision plot of the profiles for this PhotoRag paper



That's the same display as above - the lower line shows the i1 profile w/o BPC with a black level of L~14,
the middle line the i1 profile with BPC active and the third line the Spyder profile.
If you now take a gray spot with a luminance of L=40 you'll get the patch printed
- with L=40 w/o BPC
- with L=42 with BPC on
- with L=45 with the Spyder profile , that's what this B/W input/output diagram easily shows you.

Robert asked the Datacolor support line why his prints are getting lighter with the Spyder profile but more correct with the i1 profile.
The answer took a while but was pretty short - they just told him that he cannot print RAL colors with a Spyder profile and that he should use the saturation rendering intent - without any further explanation or reference to
technical documents explaining what is going on.
This was quite a disappointing answer, but it is apparent from the above diagram that you cannot print those RAL - or other - colors correctly with a SpyderPrint profile , at least not with the correct lightnesss. And there is no option in the Spyder software to get around this. Raising the lightness in the darker/shadow areas helps to reveil more details there, which is the intention and benefit of the BPC or Spyder approach , and that's just fine for regular image printing, but can create some issues for a user with profiles of different origins.

Be aware that there is another secret in the Spyder software, you find a few references to an L-Star workflow e.g. here almost at the bottom of the option list for the Spyder5 Elite version

http://www.northlight-images.co.uk/datacolor-spyder5elite-review/

But the review does not give any details about this option, an L-Star option is as well available in the SpyderPrint software without any explanation what it does. Robert did a test and created profiles with and w/o this L-Star option - he could not see a difference in printouts , and gamut plots did not show any differences either.

You won't find much info elsewhere about this L-Star option but just a reference to some patents

L-Star technology is Licensed Property of INTEGRATED COLOR SOLUTIONS INC., Patent No.: 7,075,552 and No. 6,937,249. ©2007 Datacolor. Datacolor and other Datacolor product trademarks are the property of Datacolor

So much about the Datacolor SpyderPrint profiling software

Since I plan to aquire an instrument for profiling at some future time I read this thread with much interest.

Ink stained Fingers said:

Robert did another target scan - by spot reading which takes quite some more time - that finally resulted in a profile which was almost identical with the one I created with the X-Rite i1 software.

Click to expand...

I cannot combine this statement with the two profiles in the Gamutvision plot above:

Ink stained Fingers said:

- with L=42 with BPC on
- with L=45 with the Spyder profile , that's what this B/W input/output diagram easily shows you.

Click to expand...

In this plot the profiles seem not identical for me. Is the agreement better for color printing?

I nearly always print with relative colormetric intent without BPC. This seems not to be possible with the Spyder print.