making your own targets to profile a scanner

[Phab Guy]

From my reading and understanding of various posts and other documents, to properly profile a scanner for a certain input medium, one needs to use a target produced on the same or very similar medium. Since the standard IT8-type targets are printed on film emulsions and photo papers, the profiles they produce will work well to scan transparencies or photo prints. But not for scanning, say, an acrylic or watercolour painting. Indeed, certain saturated pigments, such as in a painting, undergo significant colour shifts when scanned.

Since I scan a lot of paintings, I have created some "targets" with acrylic paints that I use to help with my colour corrections. I scan the "targets" and then take spot samples with a ColorMunki Photo and use the spot samples to help create adjustment layers in Photoshop to compensate for the colour shifting in the scans. The ColorMunki produces Lab values and I use Lab color samples in Photoshop to make the adjustments.

But that's a lot of work and is only a good starting point for doing the corrections. Instead, is it possible to create a bona fide 'IT8-like" target using various acrylic paints, measure all the patches with the ColorMunki and use those Lab measurements to make an actual funtioning profile of the scanner for the acrylic medium?

If so, can anyone point me to some documentation describing the format of the IT8 target reference file?

Or maybe this idea is just stupid?

Thanks.

Cheers!

Brad

[André Dumas]

From my reading and understanding of various posts and other documents, to properly profile a scanner for a certain input medium, one needs to use a target produced on the same or very similar medium. Since the standard IT8-type targets are printed on film emulsions and photo papers, the profiles they produce will work well to scan transparencies or photo prints. But not for scanning, say, an acrylic or watercolour painting. Indeed, certain saturated pigments, such as in a painting, undergo significant colour shifts when scanned.

Since I scan a lot of paintings, I have created some "targets" with acrylic paints that I use to help with my colour corrections. I scan the "targets" and then take spot samples with a ColorMunki Photo and use the spot samples to help create adjustment layers in Photoshop to compensate for the colour shifting in the scans. The ColorMunki produces Lab values and I use Lab color samples in Photoshop to make the adjustments.

But that's a lot of work and is only a good starting point for doing the corrections. Instead, is it possible to create a bona fide 'IT8-like" target using various acrylic paints, measure all the patches with the ColorMunki and use those Lab measurements to make an actual funtioning profile of the scanner for the acrylic medium?

If so, can anyone point me to some documentation describing the format of the IT8 target reference file?

Or maybe this idea is just stupid?

Thanks.

Cheers!

Brad

Hi Brad,

Not stupid at all, but....


First, if you go to the Kodak's website () you will find several reference and information files of the IT8.7/2, with that you should be able to see how the file is formatted and I also suggest that you read Don Hutcheson’s document “RGB Scanning For Color Management”  () and “Color Error in Digital Imaging for Fine Art Reproduction” by Robin D. Myers () .  Here are the links:

ftp://ftp.kodak.com/gastds/q60data/TDF_FILES/
http://www.hutchcolor.com/PDF/Scanning_Guide.pdf
http://www.betterlight.com/downloads/confe...Repro_color.pdf.

Regarding  “creating a…target using various acrylic paints” :  In my **personal experience** I have found that only a very few artists use mostly the same pigments and mostly the same mixtures in most of their paintings, but in those cases when they do, and *if* they do mostly small format paintings, then building a target specifically for those artists is worthwhile.  I have done some of those for watercolorists and it works but it hasn’t been a worthwhile experience as I explain below.

My suggestion to you is to forget about using actual pigments to build your targets, artists use pigments in mixtures that vary from painting to painting and many of them work with different pigments from different sources depending on availability and prices.  I have created several profiles from targets that I built from watercolor patches used as reference files and I haven’t found them as good (in the long run) as the reference file that I use now.

Now I no longer *scan* watercolor paintings even scanner-size ones, I use a digital camera to capture the art and print them using my own printer profile built from a 1785-Patch reference file from X-Rite.  Regarding *acrylic* paintings, I never had the occasion of scanning those  because none of the artists that I work with do any scanner-size paintings, again with those I use a digital camera.

I imagine that you use a  9x12 or 10x14  scanner and like me you probably scan larger paintings in 2, 3 or 4 sections ?  If so, consider that sooner or later you will probably  have to reproduce large paintings, too large for the scanner and then you will have to develop an expertise in photographing those paintings, and when you do, my guess is that you will never go back to scanning.  That is how it went with me but of course your experience and abilities might prove me wrong!    


André Dumas

[Thomas Krüger]

Mr. Faust from www.Coloraid.de creates various IT8 targets for scanners: http://www.targets.coloraid.de/

[Phab Guy]

Hi Brad,

Not stupid at all, but....

Thanks...I needed that!

First, if you go to the Kodak's website () you will find several reference and information files of the IT8.7/2, with that you should be able to see how the file is formatted and I also suggest that you read Don Hutcheson’s document “RGB Scanning For Color Management”  () and “Color Error in Digital Imaging for Fine Art Reproduction” by Robin D. Myers () .  Here are the links:

Interesting that you mentioned Robin Myers paper. His work on comparing the colour accuracy of different profiling programs used a custom test image that he made from paints. It showed noticeable colour shifting and differences between the three programs when profiled using conventional photography-oriented IT8 targets. That's what got me thinking about using targets made from real paints, not photo paper.

...
Now I no longer *scan* watercolor paintings even scanner-size ones, I use a digital camera to capture the art and print them using my own printer profile built from a 1785-Patch reference file from X-Rite.  Regarding *acrylic* paintings, I never had the occasion of scanning those  because none of the artists that I work with do any scanner-size paintings, again with those I use a digital camera.

I imagine that you use a  9x12 or 10x14  scanner and like me you probably scan larger paintings in 2, 3 or 4 sections ?  If so, consider that sooner or later you will probably  have to reproduce large paintings, too large for the scanner and then you will have to develop an expertise in photographing those paintings, and when you do, my guess is that you will never go back to scanning.  That is how it went with me but of course your experience and abilities might prove me wrong!  

I have done some very large originals with my scanners. I have two A4-size and one tabloid-size scanner and have successfully handled 48" x 48" pieces (about as large as I could reasonably handle and that customers have). I have build custom scanning tables to help me.

On occasion I have used a digital camera (when the relief of the original was too large for the DOF of the scanner). I have also considered getting a Better Light scanning back. But it seems to me there are just as many issues with the camera approach as with the scanner approach, including the setup space requirements, the capital cost, getting the lighting just right, the lower capture resolution (unless multiple shots are taken like with the scanner) and the fact that the colour response/correction issue still remains. Robin Myers findings used a scanning back and demonstrated the colour issue quite well.

So I'm currently still on the scanner road and looking to eke the most out of it. I find most of my time is spent in the colour correction stage so that's where I'm looking to improve matters.

Cheers!

Brad

[degrub]

Hi Brad,

You may want to contact the people at FLAAR

http://www.flatbed-scanner-review.org/flat...drum_links.html

They appear to deal with some of the issues you are concerned with.

P12 of the article  -

http://www.wide-format-printers.org/FLAAR_...USE_Scanner.pdf

where they mention the BARBIERI spectrophotometer for similar application.

Frank Worley

[André Dumas]

Thanks...I needed that!

Interesting that you mentioned Robin Myers paper. His work on comparing the colour accuracy of different profiling programs used a custom test image that he made from paints. It showed noticeable colour shifting and differences between the three programs when profiled using conventional photography-oriented IT8 targets. That's what got me thinking about using targets made from real paints, not photo paper.

I have done some very large originals with my scanners. I have two A4-size and one tabloid-size scanner and have successfully handled 48" x 48" pieces (about as large as I could reasonably handle and that customers have). I have build custom scanning tables to help me.

On occasion I have used a digital camera (when the relief of the original was too large for the DOF of the scanner). I have also considered getting a Better Light scanning back. But it seems to me there are just as many issues with the camera approach as with the scanner approach, including the setup space requirements, the capital cost, getting the lighting just right, the lower capture resolution (unless multiple shots are taken like with the scanner) and the fact that the colour response/correction issue still remains. Robin Myers findings used a scanning back and demonstrated the colour issue quite well.

So I'm currently still on the scanner road and looking to eke the most out of it. I find most of my time is spent in the colour correction stage so that's where I'm looking to improve matters.

Cheers!

Brad

Yes an 11X17 scanner on a custom scanning table, I can see the possibilities (as long as the art is not already framed).  I also, could have gone that way.  But when I was scanning, putting all the scanned pieces together correctly was also a problem, took a lot of time and it wasn’t perfect.  I had to cheat a bit. Whichever way you go, you will still need to correct the composite, but maybe less so with the scanner than with a camera, I admit.

After creating the target, how will you proceed in order to create the profile ?

Andre Dumas

[Phab Guy]

P12 of the article  -

http://www.wide-format-printers.org/FLAAR_...USE_Scanner.pdf

where they mention the BARBIERI spectrophotometer for similar application.

Frank Worley

Thanks Frank, good tip. I will be sure to read that article.

Cheers!

Brad

[Phab Guy]

Yes an 11X17 scanner on a custom scanning table, I can see the possibilities (as long as the art is not already framed).  I also, could have gone that way.  But when I was scanning, putting all the scanned pieces together correctly was also a problem, took a lot of time and it wasn’t perfect.  I had to cheat a bit. Whichever way you go, you will still need to correct the composite, but maybe less so with the scanner than with a camera, I admit.

I use Photoshop CS5 (and previous versions) to do the stitching and it works very well. My scanning table has an alignment grid on it that makes it fairly easy to move the original artwork in uniform rectilinear steps with uniform overlap for stitching. This part of the process, while tedious, is not challenging and seldom gives me any grief. And I get a real 300 dpi (or more) across the entire piece.

After creating the target, how will you proceed in order to create the profile ?

I plan to measure the target with the only tool I have (Colormunki Photo) and make a reference file. Then I have several pieces of software that can make profiles: SilverFast Ai IT8, VueScan and Monaco EZColor. The knowledge gap for me is converting the Lab readings from the Colormunki into the IT8.7/2 data format. Thanks for your link to the Kodak site---I have started to read their files.

I'm also told that the XRite Color Checker SG target produces profiles that work well for artwork. However, the reference file is not included for the $300 price tag, which is a stumbling block. I think I read somewhere that you can download the trial version of ProfileMaker and pull the reference file from it. So I might give that a go.

However, I'm intrigued by the possibility of making my own targets representing custom colour spaces. I have many repeat customers and I could conceivably make a target tailored to each artist's paint set. If this moves me closer to the correct colour right in the scan, then I would spend much less time in post-processing and I could turn around more work in less time.

Cheers!

Brad

[André Dumas]

From my reading and understanding of various posts and other documents, to properly profile a scanner for a certain input medium, one needs to use a target produced on the same or very similar medium. Since the standard IT8-type targets are printed on film emulsions and photo papers, the profiles they produce will work well to scan transparencies or photo prints. But not for scanning, say, an acrylic or watercolour painting. Indeed, certain saturated pigments, such as in a painting, undergo significant colour shifts when scanned.

Since I scan a lot of paintings, I have created some "targets" with acrylic paints that I use to help with my colour corrections. I scan the "targets" and then take spot samples with a ColorMunki Photo and use the spot samples to help create adjustment layers in Photoshop to compensate for the colour shifting in the scans. The ColorMunki produces Lab values and I use Lab color samples in Photoshop to make the adjustments.

But that's a lot of work and is only a good starting point for doing the corrections. Instead, is it possible to create a bona fide 'IT8-like" target using various acrylic paints, measure all the patches with the ColorMunki and use those Lab measurements to make an actual funtioning profile of the scanner for the acrylic medium?

If so, can anyone point me to some documentation describing the format of the IT8 target reference file?

Or maybe this idea is just stupid?

Thanks.

Cheers!

Brad

[André Dumas]

Hi Brad,

This PDF from Robin D. Myers is about photographing and watercolor but some of it also applies to scanning in general:

http://betterlight.com/downloads/whitePape...urate_photo.pdf


André Dumas

[Scott Martin]

Is it possible to create a bona fide 'IT8-like" target using various acrylic paints, measure all the patches with the ColorMunki and use those Lab measurements to make an actual funtioning profile of the scanner for the acrylic medium? Or maybe this idea is just stupid?

Not a stupid idea! I agree that film based targets are less than optimal for fine art reproduction work. I've gone down a very, very long and exhausting road making my own inkjet based targets on a variety of media, creating references files for them, scanner profiles from them and compared the final results. And comparing those to results from profiles made with a variety of commercially available targets.

The bottom line that I concluded is that the ColorChecker SG is the best target on the market for making ICC scanner profiles. For most situations PMP's default size, paper gray, and GTI D50 settings are excellent.

PMP and MP's default reference file works surprisingly well for the CCSG but one can make their own reference file by measuring their own CGSG - that shouldn't be a deal breaker. The difference is slight but one that super geeks will appreciate. Brad, I'd recommend getting the Colorchecker SG and skip going down the custom 'IT8-like' target rabbit hole!