Mathematically Modeling Light Sources

[brentwing]

I am a colorist for a small company that prints color correct graphics for clients.  Each of our clients is allowed to specify the type of light source that they will use to evaluate our prints.  The client provides a physical sample of a color which is then custom matched to their lighting "standard."  We match a color to an RGB value and enter the value into a database.  However, we recently have started to match prints to multiple light sources, requiring us to match the same color multiple times.

In an effort to streamline our color matching process, I'm trying to create a mathematical profile for each light source (D65, D50, CW, etc.).  My goal is to have a conversion model that allows me to specify an RGB value and a light source and have the formula return the RGB value needed to print a color correct image.  Using a Photoshop curve adjustment isn't going to cut it.

If anyone has any ideas, or has done/read about something similar, please let me know.  Thanks in advance.

[opgr]

Reading your post literally you would be looking for information about "Chromatic Adaptation" and your best bet would be the Chromatic Adaptation Calculator as found on Bruce Lindbloom's site.

However, printed output can exhibit colorshifts under different lightsources that are not solely related to chromatic difference between those lightsources.

You should be aware that our perception is very sensitive to areas of uniform color (for photographers: that includes B&W images). If you're referring to colorsamples for textile or clothing, or carpaint for example, you might well run into the colorshift problem mentioned and you may not be able to solve the problem by simple Chromatic Adaptation calculations. You would then need a custom table as you are already building, and even then the colordifferences may be beyond the precision of current colormanagement implementations and/or output devices.

In an effort to streamline our color matching process, I'm trying to create a mathematical profile for each light source (D65, D50, CW, etc.).  My goal is to have a conversion model that allows me to specify an RGB value and a light source and have the formula return the RGB value needed to print a color correct image.  Using a Photoshop curve adjustment isn't going to cut it.

If anyone has any ideas, or has done/read about something similar, please let me know.  Thanks in advance.

[colourperfect]

May I ask why you dont match LAB values measured with a spectrophotometer ?

Ian

http://profiles.colourperfect.co.uk

[digitaldog]

Do you have something like an Eye-One Spectrophotometer that allows you to measure the illuminant? If so you could build profiles using that data in ProfileMaker Pro instead of using a default D50 assumption.

[brentwing]

Unfortunately, our colors are often matched to textile samples, so colorshift differences have been an issue.

We do have a spectrophotometer that has been used to record the illuminant, mostly to track discrepencies between different brands of bulbs.  I don't use the spectrophotometer myself and I'm unfamiliar with ProfileMaker Pro.  I might be incorrect in my thinking, but couldn't this cause blowouts in certain sections of the color spectrum?

I appreciate all the comments and will look into all that you have suggested.

[brentwing]

BTW

If anyone else has anything to add to this discussion, I'd greatly appreciate your input.

Thanks