DIY camera color profiling target f/ Argyll CMS

[ComputerDork]

This is going to sound like a stupid question to lots of people since it's not worth the effort except as an educational personal science project maybe, but I was messing with using Argyll CMS this weekend to create ICC profiles for camera calibration, and the ability to create a definitions for custom charts, combined with the fact that all I have is a ColorChecker, got me to thinking.

Unless I'm missing something, it seems like I should be able to pick some useful range of colors, print a chart with say 300 patches on my printer without regard to any existing target standards, create a CHT file for the layout of my chart, then measure all of the patches manually with a spectrophotometer to create the CIE file containing all of the Lab measured values. Assuming I pick good colors and I have a printer that can print them (the really dark ones sound the most difficult), it seems like I ought to be able to use the chart to create a decent LUT-based camera profile.

Given that I have an Epson R3000 (UltraChrome K3, etc) and a ColorMunki Photo, what would the issues be if I wanted to create my own personal digital camera color profiling target like this?

I understand only the most basic issues with limited numbers of pigments, metamerism, lack of diversity in spectral reflectance distribution of patches, gamut limits of the inks involved, and the amount of work required to measure 300 or so patches with something like a ColorMunki that doesn't really have a sensor aperture that's suited to reading lots of little patches, and stuff like that. But I don't know exactly how to quantify how useful or useless a chart printed with something like UltraChrome K3 pigments would be in terms of pigment diversity or gamut range for the purpose of profiling a digital camera under various lighting conditions.

Even if it works reasonably well it's not likely to be the greatest value for the effort compared to just buying a target, but it might be worth doing as a self educational science project.

[Ernst Dinkla]

Your basic pigments in the printed color chart will not be more than 5, 6 or 7. Most commercial cards exceed that number already.

A good RAL swatch guide of car paints, house paints or offset or silkscreen ink catalog swatches or artists acrylic paint swatch books have more pigment samples. A good UV enabled and a UV cut spectrometer to measure the colors is a next step and allows you to see whether the pigments actually differ in their spectral plots, you might be able to see whether the orange sample is a genuine pigment or based on a mix of yellow and warm red. It is also a good thing to keep fade properties of the patches in view, both pigment and original paint medium + the substrate characteristics: does it contain OBAs, is the paper or plastic shifting color in time. The surface should be matte or satin at most. Sericol or Marabu silkscreen basic mixing ink swatches exist, I think the pigment number can be up to 10.


--
Met vriendelijke groet, Ernst

340+ paper white spectral plots:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
update april 2012: Harman by Hahnemühle, Innova IFA45 and more

[ComputerDork]

Thanks Earnst. I think I'll go out to the hardware store and get some potentially suitable paint chip strips to play with. I'm just not sure how many they'll part with. At least I have a somewhat legitimate reason for getting some as I'm going to need to paint a wall for a photographic background and will end up buying paint, so I won't feel too guilty.

The OBA issue is annoying, and I'm not sure I want to shell out for a fancy spectro without a UV cut filter just yet. (The ColorMunki has a UV cut from what I remember.) I think there may be some "poor man's spectrophotometer/colorimeter" things I can do with Argyll CMS if I can get the illuminants right, though I'm sure they won't be super accurate. Maybe that'll be good enough to at least get some sense of the impact of OBAs, if I can get the method to work.

Thanks for the work you've done in making paper spectral plots available. I always find things like that to be useful, and I'm sure more people appreciate the effort and willingness to share data than the number that actually mention it.

Not sure why I like playing with this stuff. I'm starting to think I should go for a degree in color science or something if I'm this interested in just playing around.

[Ernst Dinkla]

If I recall it correctly the Munki measures in reflective mode down to 430 NM and copies the 430NM measurements down to 380 NM so it draws a straight path in unknown territory from 430 to 380 NM. A UV-cut spectrometer does better than that.

http://lists.apple.com/archives/colorsync-users/2009/Jan/msg00104.html


--
Met vriendelijke groet, Ernst

Dinkla Grafische Techniek
Quad,piëzografie,giclée
www.pigment-print.com

[ComputerDork]

So it sounds like it's both a UV and visible V cut spectrophotometer because they went a little cheap on the UV cutting. From what I can tell though, human eyeballs aren't that sensitive to the affected range so hopefully the error compared to human perception is minor.

I went to the hardware store and they didn't seem to care how many paint chips I grabbed, so I got like 200. Wouldn't surprise me to find that they'd like to get rid of them within a certain time period anyway in case they fade. Only problem with them is that I'm not sure how many of them are approximated with some sort of process colors and how many use the actual pigments in the paint, but given that one manufacturer, at least, has a little mini multi-illuminant viewing booth in their paint chip display, I'd hope that whatever is on the chips would have very similar spectral reflectance distribution since metamerism would have to be similar to the real paint to make this useful.

I haven't messed with since grabbing a bunch of paint chip cards, but I figure when I get back to this I'll probably print up a chart design based on the range of colors used in some existing charts, then die cut paint chip cards and try to place paint chips on the printed chart where they most closely match what the printer printed. I'll have to leave some of the patches, at least, to the printer (to get the proper gray step range, etc. believe it or not people don't seem to have much demand for paint chips distributed evenly between L=0-100 in small increments). I'll have to look up which papers I have are OBA-free or closest to it. I still don't know how likely interior paints are to have OBAs or how much of them they'd have. Most people seem to like off-white colors for white anyway, so is there really a demand for them in most "white" paint even? It seems kind of likely for bright colors though.