reproducing art or other things

[sgwrx]

so, what is a general process for making a print from a captured image of artwork (painting/watercolor) or a carpet sample or tile sample.  i'm guessing it's probably similar, again in a general sense.  would it go something like this:

1. setup the artwork or other sample and light it under a specific color of light (5000k or whatever).  make sure that light is as "full spectrum" as possible.  measure that light's spectral distribution and color and it's intensity, taking test photos to make sure the light levels are even across the sample.

2. photograph the sample with as little lens distortion as possible, dead on level in all axis and dead on center. use at least a grey card to set white balance. preferably have a camera profiled.

3. bring that photo into whatever program, calibrate and profile the monitor, soft proof to get somewhat close, then make test prints and directly compare with the original sample under the same lighting conditions.

i'm way over simplifying, but is this the general idea?

i remember reading something years ago on the net about how painstaking it is and sometimes nearly impossible to match all the colors in an original artwork.  but on the other hand, i mentioned carpet sample because where i work, a manufacturer sent us a 50cmx50cm print that looked pretty darned close to an actual sample of that carpet we also had.  so maybe it wouldn't be too hard to get to a level where it's a "pretty good match"?

thanks,

[Lundberg02]

Search this forum for imaging artwork. It's hard. Very hard.

[tom b]

It's hard and it's simple.

To a trained eye it is hard.

For the vast majority of the general public it is simple. You have to calculate what your audience is.

Good luck,

[torger]

Profiling the camera is absolutely essential as bundled profiles are not designed for colorimetric accuracy. Reaching a "pretty good match" is quite simple. Making the highest end possible is complicated, involving designing custom targets and more. The difference between results of the pretty good match and the highest end possible is in most cases very very small.

On forums you often see descriptions that make things seem very difficult and complicated, almost impossible, but it's often because the workflows described try to minimize every source of error to an extreme, often what's beyond reasonable for a given customer.

[amolitor]

How much error must one really squeeze out, given the variability of observers?

That's a serious question. I don't know how the variability in viewers compares to the sizes of typical errors, and I'd like to.

[torger]

How much error must one really squeeze out, given the variability of observers?

That's a serious question. I don't know how the variability in viewers compares to the sizes of typical errors, and I'd like to.

What do you mean with variability of observers? Variability in skill of differing between small color differences, or variability in for example how you name colors, or differences concerning metamers? In this case it's only about matching colors, and as far as I know eyes with working color vision should have the same "shape of color matching functions", although it varies how skilled people are in detecting small color differences.

I haven't seen any research in this area but it would surely be interesting to read something about it.

[Iliah]

> is this the general idea?

Depends on the customer. For archival quality at least 2 additional colour filters are used (3 shots), and it may be a scanning back.

[digitaldog]

Search this forum for imaging artwork. It's hard. Very hard.

Agreed. It can be very hard, sometimes, depending on a number of factors, nearly impossible. For example, no shipping camera that I know of meets the Luther-Ives condition. This means that cameras can exhibit significant observer metamerism with respect to humans. Doesn't mean those doing this kind of work shouldn't do their due diligence in proper setup and color management, but again, depending on many factors, just the subject itself, it might be nearly impossible to produce a visual match of all colors. Client education is necessary such they understand the limitations of the technology.

[amolitor]

What do you mean with variability of observers?

I believe that it is possible, even feasible, to have an object (e.g. a painting), and two people A and B, and two prints with the following properties:

Both A and B agree that the two prints are different. Each thinks one of the prints is a better match to the painting than the other.

But they disagree about which print it is.

If indeed this is the case, then fining your process down to give reproducibility and color accuracy substantially greater than the differences between these two prints (however you choose to measure it) strikes me as futile. I have no real sense of what the scales involved here are, though, hence my original question. Is it Crazy Hard to get this close? Or is it only Moderately Hard and beyond that it's about pleasing oneself? Or what?

Something I rarely see mention of is comparing the color in the print to the actual object photographed. This is impossible with, say, a landscape, as the light will never be the same. Color Management is often reduced to the problem of making prints that look as close as possible to the picture on the screen, without reference to the original scene or original object. It is reduced to managing mappings between color spaces, without reference to the physical full spectrum colors that are, ultimately, the source of it all.

Reproducing Art is, on the other hand, ALL ABOUT the latter.

[Stephen Ray]

what is a general process for making a print from a captured image of artwork (painting/watercolor) or a carpet sample or tile sample

I think you may have left out a couple of things...

1) Determine if the original color gamut can be reproduced with the particular process.

2) You may need to manually select outlier colors via alpha masks or paths and tweak (typically using curves) accordingly. Much of commercial product photography files have these color modifications to control color across repro methods and over time. Sometimes you might notice a website’s photo of shoes is really the same photo re-worked to the different colors offered.

And yes, it's common for a commercial color house to have an actual sample of the item to travel in the job jacket and a fine art printer to have the original art or an approved proof or film chrome to use as the guide.

An interesting way to match color was apparently used by the Dutch artist Vermeer, as discovered recently by Tim Jenison and documented in Penn & Teller’s film, ”Tim’s Vermeer.”

[kirkt]

This is a pretty nice discussion of some of the issues and a proposed method for working with digital reproduction:

http://www.trumpetpower.com/photos/Exposure

kirk

[amolitor]

Anything that involves starting from RGB or CMYK output is, to some degree, cheating. You're already in the land of "well, physical spectra are hard, so let's just mix a few colors together and try to create the same response in the human eye" which is a fine thing, but it's the *end* of the process, not the *beginning* when you're talking about reproducing art.

[digitaldog]

Anything that involves starting from RGB or CMYK output is, to some degree, cheating.

As opposed to what other color model for output?

[amolitor]

Those are fine as endpoints, as I noted. I guess to first order they pretty much are the only endpoints.

The linked article was talking about a set of test patches which included inkjet output, however. Which strikes me as starting the process one step too late for the purposes being discussed here.

Any insight into the questions in #8 by the way? You seem like the guy who'd know.

[Ernst Dinkla]

Anything that involves starting from RGB or CMYK output is, to some degree, cheating. You're already in the land of "well, physical spectra are hard, so let's just mix a few colors together and try to create the same response in the human eye" which is a fine thing, but it's the *end* of the process, not the *beginning* when you're talking about reproducing art.

Multi spectral scanning and N-color printers should solve more pigment to pigment issues in reproduction. Where Multi and N can have high numbers. Just one doc on the subject:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.23.6320&rep=rep1&type=pdf

Color management targets based on way more pigment qualities would work better too in that process than they do in RGB-CMYK systems. There is limit where more patches do not work in the RGB/CMYK system if the distinction between several pigments is lost in the RGB capture and one light source. In Multi spectral scanning metamerism becomes your friend to separate pigments and define their color path in the reproduction process.

Image Engineering .de used to have an article on the HP G4010/4050 scanner where the use of two CCFL light sources, two scan runs and just one RGB linear CCD mimicked multi spectral scanning. Improved reproduction of an acrylic paint target compared to other desktop scanners that aim more at reproducing photo chromogenic CMY colorants. There were less differences between the scanners in the photo reproduction targets.

Given equal conditions in lighting and print media matched to painting substrate (OBA content mainly) one arrives at a point where two observers will both select the same reproduction as a match for the original but still have another image in their minds of both reproduction and original and you will need other experts to get that equal if needed :-)

All the above goes out of the window when nature has to be reproduced, at 5000 km distance or nearby. Color comes in many ways not just by light falling on pigments or dyes.

Horses for courses, I do not have the customers for multi spectral products.


Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2014 update, 700+ inkjet media white spectral plots

[amolitor]

Nice paper, Ernst, at least the background material they give. I've no qualifications to judge the paper's original content, but the background filled in some details for me. Thanks!

I had no idea that the paint matching guys jump straight to spectral matching. No wonder it works so well

Has anyone considered a digital toolchain that uses more than three colors? Most inputs sources are three colors, but apparently not all, and we seem to spend a lot of effort struggling inside three color systems and trying not to lose stuff. It seems like benefits similar to using an especially wide-gamut 3 color system might be obtained by translating immediately to a 16 color, or more, space, and working within something that more closely approximates spectral matching.

[Tim Lookingbill]

There is also the color perception issues vs reproduction accuracy by the numbers dealing with lights artwork is captured with versus lights the print and original art is viewed under that hasn't been addressed. There are desirable fluorescing effects from gallery lights on certain paints that cameras and camera profiles can't reproduce.

There was a local art gallery owner I was consulting with years back who was having trouble reproducing eye popping globs of cadmium yellows, oranges and red paint in southwest abstracts on canvas with his inkjet and photographing the originals with his Olympus DSLR under halogen styled CFLs. These CFLs were the same ones lighting the original artwork in his gallery and provided this fluorescing effect but I didn't know it at the time. The gallery owner even tried direct daylight from his storefront display windows and using flash and camera profiling. Nothing worked.

He ended up editing the captures using the CFLs since this was the most consistent continuous lighting he had available he could visually match too. An upgrade to an HP Z3100 wide format inkjet with constant calibration and wider color gamut helped out quite a bit as well. He showed me the originals to the inkjet canvases and they were very close and when viewing under his halogen CFLs delivered the fluorescing he desired. After several of these reproductions using this method he got up to speed and got quite a lot of work done. He sold the reproductions to the artists who created the originals.

I've provided some examples showing this fluorescing effects and lack there of from various lights including full spectrum daylight to 2800K CFLs on green leaves. The second image is how my DSLR did not capture this fluorescing shooting Raw and how I had to use my memory to put it back in. It requires careful study of how luminance is tied to hue/saturation but only through perception. The numbers aren't going to show this.