On Jul 4, 2013, at 8:38 AM, Ben Goren <ben@xxxxxxxxxxxxxxxx> wrote: > He does this by treating the input as a camera with more than just the usual > three RGB channels, and by doing the spectral math to calculate tristimulus > values from the six- or nine-channel input (if using one or two different > filtered images in addition to the unfiltered one). I should hasten to add: Dr. Berns doesn't just calculate tristiumulus values; he calculates, per pixel, a spectrographic plot. Logically, his setup is equivalent to a spectroradiograph with very high spatial frequency and low (but very accurate) spectral frequency. Getting the tristimulus values from that, for any given illuminant (real or virtual), is obviously not a challenge at that point. He's also done work to combine camera and spectroradiograph readings to merge the high spatial frequency of the one with the high spectral frequency of the other. I'm pretty sure that's today's state of the art, but I wouldn't at all be surprised if he's moved on to even more impressive stuff. Cheers, b&