Gamut of custom profile much smaller than Epson profile?

[AndyF]

A problem with some colours on a P800 caused me to make a custom profile, and the print is better now but the new profile has a much smaller volume than the standard Epson icc profile.  I'm wondering if that means the custom profile is not as good as it should be and not all the colours that could be printed on that printer/paper/ink are available now?  I would expect the Epson one to be within a few percent of what it should be, so that's my reference.

The normal profile is Epson SC-P800 Series Premium Photo Paper Glossy.
The custom profile was made using Silverfast 9's printer cal tool, after first calibrating the scanner itself with an enhanced IT8 target.
The iccview.de website and GamutVision do have slightly different results but in one case the custom profile is 67% the volume of Epson's and in the other it's about 88% of Epson's.
Attached are views of those two results.
Thanks for any input on this, or experience with Silverfast 9 making icc profiles.

[digitaldog]

Make prints with color reference files designed for this such as
http://www.digitaldog.net/files/Gamut_Test_File_Flat.tif

Next see this about how not all gamut mapping is equal:
http://www.colorwiki.com/wiki/Color_Management_Myths_26-28#Myth_26

The proof is in the print.

[Rhossydd]

Thanks for any input on this, or experience with Silverfast 9 making icc profiles.

Trying to make printer profiles with scanners is far from ideal. It's far, FAR better to use a spectrophotometer and dedicated profiling software.

Just get one of the small custom profiling companies to build you a profile with decent kit, it will save you a lot of time and stress and really won't cost much.

[Simon J.A. Simpson]

Trying to make printer profiles with scanners is far from ideal. It's far, FAR better to use a spectrophotometer and dedicated profiling software.

Just get one of the small custom profiling companies to build you a profile with decent kit, it will save you a lot of time and stress and really won't cost much.

+1.  The intervention of the scanner’s restricted gamut will almost certainly introduce a reduction in the measured printer gamut.  Far better to use a dedicated spectrophotometer or a reputable profile making service.

[digitaldog]

Trying to make printer profiles with scanners is far from ideal. It's far, FAR better to use a spectrophotometer and dedicated profiling software.

Just get one of the small custom profiling companies to build you a profile with decent kit, it will save you a lot of time and stress and really won't cost much.

Indeed and the last time I tried, using the same product as the OP, the profile was awful.

[Rhossydd]

using the same product as the OP, the profile was awful.

It's surprising that Silverfast continue to include this dodgy gimmick feature in their otherwise well respected product.

[digitaldog]

It's surprising that Silverfast continue to include this dodgy gimmick feature in their otherwise well respected product.

Indeed, and they asked me to test this for them, I did, reported back the awful results, like two years ago, crickets. The profile produces was absolutely unsuable. Not even close. 

[AndyF]

Thank you all for the comments on the SF profile!  I might ask them for a refund on that license since it's poor, although the scanner cal license was useful.  For the V550 scanner, the average  delta-e was 0.6 on an 856 patch target, just as info.

I'll look for someone or a service to make a custom profile for me.  It will then be interesting to how well the photo prints, and how the profile compares to Epson's version.

One problem with printer test files is the lack of a physical reference on what they should look like.  It might print with a smooth gradient but not the particular shade of colour that it should have.  To partly solve that I've ordered my print, a colorchecker grid, and the Outback test photo printed from two separate pro services who will print the files "as-is", no adjustments.  Of course their printer and inks are likely different than mine, but still useful.

[Doug Gray]

Thank you all for the comments on the SF profile!  I might ask them for a refund on that license since it's poor, although the scanner cal license was useful.  For the V550 scanner, the average  delta-e was 0.6 on an 856 patch target, just as info.

dE's when making profiles are self referential and pretty much meaningless even with spectros. Evaluating profile accuracy requires measuring a set of colors that are independent of the set used for making a profile.

One problem with printer test files is the lack of a physical reference on what they should look like.  It might print with a smooth gradient but not the particular shade of colour that it should have.  To partly solve that I've ordered my print, a colorchecker grid, and the Outback test photo printed from two separate pro services who will print the files "as-is", no adjustments.  Of course their printer and inks are likely different than mine, but still useful.

It would be great if there was a place that one could buy a reference image that was in-gamut and a print using relative colorimetric that could be compared against one's own printing. Not aware of any.

Printing a reference image, for instance from babelcolor, using absolute colorimetric should be a very close match to an actual colorchecker card. Especially for matte prints on paper that have no OBAs. But this won't tell you much about how the profile maps colors that are near or out of gamut using perceptual.

[markgunion]

AndyF,  if you're looking for custom profiles you should check   www.digitaldog.net .    Yes, that's the same Andrew Rodney that's been replying to your posts.   I have two custom profiles recently made by him for my Epson P800.    I'm pleased with the profiles, the price was reasonable, the process was simple, and Andrew was quite prompt.   Recommended.

[AndyF]

Hi Mark
Thanks for the recommendation.  I had found two other sources (Great Printer Profiles, and Profiles by Rick) and printed their charts but not placed the orders or mailed them yet.  For some reason digitaldog wasn't in the first couple of pages of Google results (I see it now, third page).  Glad you pointed out his service, I'll order a profile from digitaldog (I already have your book, Andrew!).
Andy