very specific problem from a biologist (yeah, biologist)

[qosta]

hello there!
I do no not work with printing per se. I'm a biologist in my master's and part of my research involves printing butterfly wings.
Mind that my research is not the printing itself, but what I'll do when I'm done with the printing: I've 'read' butterflies' wings in a spectrophotometer, have a profile (a mean curve if you will) to use as a pattern and have printed wings that will, later, be read as were those real wings. We're creating fake butterflies so we can use them in our experiment (the strangest that it may sound, it's a vastly used method)

the thing is, I've have already taken all the measures and have a few discrepancies in my curves (my 'real' profile and my 'fake' profile. (so to have a better grasp: my curves are in nm, my 'fake butterflies' curves are slightly more reflective than they should from 550nm to 250nm. I'd want the curve to have a slightly more abrupt descent at around 550nm than it does). What could it be the problem I'm facing? Is there a way to feed a image edition program a nm curve and have it printed? Now I'm using Photoshop. Should I be using Corel Draw? Am I absolutely wrong? Any answer is welcome.

My best regards

 

[Stephen Marsh]

What spectrophotometer are you using to measure the real butterfly wings? My guess is that you would need a spherical based instrument.

Keep in mind that the source colour may not be reproducible in print.

If I remember my Nat Geo/Discovery correctly, butterfly wings are iridescent and don’t have a “true colour” (light angle, viewing angle etc), so this is hard to simulate with a single flat colour in 2D. Gradating between darker and lighter tones and coloured hues may help? You would likely have better luck using Lab (L*a*b*) values in graphics software as spectral reflectance curves are not as common.


Stephen Marsh

 

[Schnitzel]

Sorry, I didn't get everything, but from what I understand, you have a spectrum of a color that you wish to print. If that's the case, head over to Bruce Lindbloom's site, under "Math" you have "Spectrum to XYZ". The last 4 equations are probably what you need. You'll need the spectrum of the illuminant under which you'll view the color (generally D50 or D65), the spectrum of the sample (your measurements), and the spectra of a standard-observer (generally 2°). Mind you, the illuminant and standard observer are based on the visible spectrum, and it seems your measurements extend beyond that.

This will give you the CIEXYZ values for that spectrum, which then you'll need to convert to RGB. Bruce has a calculator in his website, under "Calc", "CIE Color Calculator", input the 3 values you got and press the XYZ button. Draw a patch with this RGB color in Photoshop or whatever software you use.

After that you'll need to print the color. To get a good match you'll need to use color management, which is beyond the scope of my answer here. Sorry. Good luck!

 

[gordo]

I agree with Stephen, butterfly wings get their colors from two different sources: ordinary (or pigmented) colorants and structural color. The ordinary color comes from normal pigments that absorb certain wavelengths of light and reflect others. That is the color that spectrophotometers "see" and measure.
The structural color of butterfly wings comes from the specific structure of the butterflies' wings - this is iridescence (like mother of pearl seashells). It happens when light passes through a transparent, multilayered surface and is reflected more than once. The multiple reflections compound one another and intensify colors. This is not what measurement instruments are designed to measure. So, I don't think it's possible to do what you propose.

 

[qosta]

Thanks for the answers, good to have some material to look for.

The information that butterfly wings colors can be a product of two different sources is well known and something we were aware of from the start. The color in the specimen in my research is a pigmentary color (a chemical color, if you will), not a physical/structural one. We're going to have a look at the scales pattern and structure anyways, so to have a more embracing work on scales deposition pattern, but we're very confident(from the vast literature) that this is not a case of iridescence.

Rereading my first post I understand when you say you don't understand. Here's the thing: I work with a specimen that has, roughly speaking, three colors. I've measured(spectro) a color spot in real butterflies, from that I've generated a (mean) color profile for that spot. Than I've scanned real wings, printed them and measured the printed(fake) spot as well. I want these curves to be the more similar they can be (because they're from the same spot). In the course of my trials, I''ve managed to reach a very similar curve. But I'm not satisfied with it,and I'm pretty convinced it's a printing problem.

here's an attachment: the black line is my mean reflectance for a specific real spot. the curve I want my fake spot to look like. the red line is the best fake spot I've reached for that curve so far. you can see that around 550nm the red curve should descend more abruptly. is there a way?

 

[benstarr]

The spectral reflectance curve is going to be dependent on the pigments in the inks used to print the sample. I think it's very likely that the pigments in the inks are not the same as those found in the butterfly wings. The substrate will also contribute it's color to the spectral reflectance so that needs to be taken into consideration, have you compared the curve of the substrate you're printing on without any printing to the curve you have after printing?

 

[Deleted member 16349]

Here's the thing: I work with a specimen that has, roughly speaking, three colors. I've measured(spectro) a color spot in real butterflies, from that I've generated a (mean) color profile for that spot. Than I've scanned real wings, printed them and measured the printed(fake) spot as well. I want these curves to be the more similar they can be (because they're from the same spot). In the course of my trials, I''ve managed to reach a very similar curve. But I'm not satisfied with it,and I'm pretty convinced it's a printing problem.

The source of the light from the object is of no importance. What the eye sees is just the light which hopefully is what your spectral curve describes.

I agree that it is a printing problem. You need a printing process that can match the colour. Colour is not a physical thing. There is no colour in Nature. Colour is a perception of the human eye and brain.

So generally, there are two options to get to a printed matching colour.

One is to obtain a metameric colour match. That would be a match of two colours that have different spectral curves but result in the same XY and Z tristimulus values. This is based on Colour Science. I am not a colour scientist so I won't to go into more detail because I will probably not state some issues properly.

The second option is to try to match the spectral curve, which seems to be what you are trying to do. This is a problem with conventional printing since conventional printing has limited numbers of inks used. The inks used are filters to light. Conventional printing matches colours based on the first approach that tries to obtain a metameric match.

A metameric match will only ensure that the match is made when shown in a specific light viewing condition. Put the two samples in a different illumination condition and there is probably not going to be a match. Different observers can also affect the metameric match, where in the controlled illumination, one observer will see a match while another will not.

A spectral match is the ultimate since the two samples will match in any lighting condition and for any observer. That is not saying the colours will always be the same but only that there will be a match between the two samples. This would be the case as long as there is no fluoresce involved.

The problem is that to have a printing process that will result in a matching spectral curve would require lots of different inks that could filter at very specific regions of the visible spectrum so that one could fabricate the spectral curve as desired. I don't know if there are printing processes that have the capability of obtaining the desired accuracy of curve fabrication that you would require. There are some inkjet printers that have extended ink sets that might get you closer but you would also need some software that could attempt to match the spectral curves and not the usual metameric match. I am not sure what is available. It is a question of capability of the existing technologies to do what is physically required.

There has been work with reproducing fine art based on spectral curves matching. You might have some ideas if you look in that direction. Graphic arts is not the field to be talking to but maybe the imaging sciences. Two very different groups.

*I would also add that if you see different colours from the spot on the wing, at different angles, then please be aware that the spectrophotometer which uses a fixed geometry, such as 45-0 degree, will only tend to pick up one angle. So the type of spectrophotometer is important and how you use it. If you think a spherical spectrophotometer, that collects all the light coming from the sample in all directions would be good, I think that would be wrong since it will just add all the different colours from different angles and will not be what you see.