RGB VS Multichannel ICC Profile for Fine-art Reproduction

Hello Experts,

I recently opened a fineart reproduction studio. using a cruse scanner, Hp z3200, Cannon ipf9100, Epson 9900 , i1 pro spectrophotometer , and of course a NEC 30’ calibrated screen and the Ergosoft studio print 14 rip software.
The problem is as follow i am having a very bad prints on canvas by bad i mean no sharpness in colors, very Dull Colors and specially a very nonsaturated red color plus a loss in highlights and shadows.
The process is as follows:
i scan on the cruse scanner along with the X rite SG color checker, ( Color management is disabled on the scanner) i do the white balance on the scanner histogram by setting whites to 235,grey to 100, and blacks to 25( average for RGB values).
then i move the target to xrite profile builder , i create a scanner profile using d 50 lighting, then i assign the profile to the scanned painting, convert to adobe RGB using relative colorimetric with Black point compensation and save it.
now for the printer profiles i am creating Halftone or 12C profiles whether on cannon Hp or Epson which i assumed they are multichannel profiles.
i do density linearization, separate ink limit, total ink limit ( 240 most of the times where no bleeding or pooling nor bronzing is available) and finally create an ICC profile with GCR 2 ( black in dark tones) and print around 1000 patches and measure them on the i1 pro.
generate profile and save it.
i am printing the image on canvas Breathing color and Innova matte canvas and i am getting a huge shift in the colors and loss of highlights and shadows.
the profiles seem to work perfectly with photo images or downloaded images. but never from a scanned image.
Today i created an RGB profile on Ergosoft and i got an almost perfect result but i didnt understand why?!
i know maybe it is a little advanced question and i was hoping if someone can help clear things up because i think using multichannel profiles and setting density values and total ink limit can be more useful.


You have a lot of variables in your system, and it might help if we can confirm what works and what is not working so that we can make troubleshooting easier.

When you bring your scanned image into PS and assign the scanner profile, does your image look correct on your display?

If you were to softproof your image in PS (using View > Proof Setup > Custom) does it look identical to what gets printed?

Are you saying that your system is working fine, printing great color except when you print on your canvas media?

Or are you saying that your system is giving you great prints except when using scanned images?

Are you saying that an RGB Ergosoft profile is now giving you perfect prints even with scanned originals printed on canvas?

A few notes:

  • You are probably not wanting to make true multichannel profiles, even if your printers have 12 inks. Most RIPs will require a typical CMYK profile, and the printing engine automatically distributes the ink to the 12 channels.
  • Canvas material is notoriously hard to profile accurately so that shadows come out with detail.
  • RIPs can be pretty tricky to setup. Make sure you’re doing everything accurately. The fact that your RGB profile works fine can help us narrow where the problem lies.

Hello Patrick and thank you for your answer!

[color=red]thanks again

Well it sounds like the scanning part of your workflow is mostly working okay.

In looking at the printing part of your workflow, I’m going back to this idea:

Canvas is hard to profile with normal measurement instruments like the i1Pro. The uneven surface of the canvas can cause the light to scatter in different directions. It will give you a lighter measurement, and it might tend toward a color bias as you’ve described. What we do in our profiling services is scan canvas targets with one of our SpectraScan tables using a polarizing filter. This filter eliminates some of the stray light coming through the measurement lens, and we find that profiles for canvas and matte material have noticeably greater detail in the shadow area. The SpectraScan tables are no longer being made anymore, but Barbieri makes a table that can use a polarizing filter. I don’t know of anything you could do with an i1Pro to get the same results. You might be able to reduce the shift in red if you take several measurements and average all of them.

You could also see if Breathing Color or Epson have canned profiles that would work on these printers on this media. Usually the media manufacturers have better equipment for making better profiles. While they would not be as good as custom profiles in some ways, maybe they would be better in others?

I have a Gretamcbeth spectrascan i think i will send it to xrite for calibration and then start using it again on canvas.


Patrick is right on. If you can build profiles for canvas material using a GMB SpectroScan XY Table with a P filter you will see better results. I have had this experience on several occasions.


Hello Marc,

Thanks for the answer. i have GMB spectro scan that i bought from ebay, but unfortunately it needs an adapter and re calibration which will cost me around 500$ to ship it to be calibrated and bring it back. so i decided to keep as a decorative item :slight_smile:

Well I’m very familiar with that RIP, and using if for N channel, it truly is N channel… and each and every color has to be linearized and limited, and light ink/dark ink crossovers, etc etc…
It’s not for the faint of heart, and often using the supplied RGB drivers and working that way is more viable.
But I’m not famliar with working with canvas so maybe not of much help.
Are you using the GPS profiler or a 3rd party profiler?

Hello Tyler, yes i am using Color GPS to create a profile.
I am now Creating RGB profiles for all kind of paper , and the color matching result is much better than what i used to get in Multichannel Profile.
i dont know what i am doing wrong in multichannel profiles, Ergosoft has a lot of variable and thats why i dont know what is the wrong step i am doing.
specially the profile in canvas.

I have a feeling my input will quickly become useless because of my lack of canvas experience, which seems has it’s own set of issues. But let me ask a few more questions-
Are the RGB profiles that are working well also made with the GPS profiler?
You mention 3 different printers, each of those have their own RIP setup complexities… are these problems consistent with all 3? Or if you are working with one, and these posts relate to that one, which is it?

yes all RGB and Multichannels are made on color Gps and the result is the same on all 3 printers ut much worse on the iPF 9100 cannon.
usually the cannon total ink limit chart is always bleeding or bronzing and i have to reduce the density a lot to be able to avoid bleeding in the chart.

Well I’m slightly at a bit of a loss then. I’d just add that due to the complexities of multichannel linearization, individual and total ink limiting, black build, etc etc… I’ve decided to go back to rgb for most standard media. I use multichannel for it’s ink control with unusual media and B&W.
I’d say put the effort into maxing out RGB setups, then going back to multichannel experimentation when you have time. But I can’t tell you anything about it for canvass, which I suspect is the issue here, though that it’s fine in RGB is slightly odd…
WIth the 9900, selecting stocahstic, prefer small dots, default light ink / dark ink dor percetages, full 100% black generation in GPS, is very workable right off the bat with standard inkjet media… tweaking beyond that can both improve things, and make things much worse… but that’s a place to start. For Epson though, it’s hard to beat their RGB screening for most media.
Sorry I can’t be more helpful.

Not at all tyler thanks a lot for the help! I believe its something that needs lots of experiments.