jon@oddhack.UUCP (04/16/87)
I need to convert the spectral types and magnitudes specified in
the Yale Bright Star catalog into RGB triplets suitable for display on
a color monitor. I would appreciate any algorithms, references, or
suggestions (in that order) on how to do this.
Thanks,
-- Jon Leech (jon@csvax.caltech.edu || ...seismo!cit-vax!jon)
Caltech Computer Science Graphics Group
__@/phil@brl-smoke.ARPA (Phil Dykstra <phil>) (04/20/87)
Here's the method I used: The Yale catalog gives the B-V color index, I. I converted these to color temperatures: Tc = 7200 / (I + 0.64) in degrees K and then integrated Planck's black body radiation curve for several temperatures against the CIE standard observer color matching curves. I converted these to xy chromaticity coordinates. Here's a table of some values with a set of equal intensity RGB value for a "typical" monitor [Note that in order to display a color "correctly" one needs to know the gamma and phosphor chromaticities of their display device and convert xy to match. These make a BIG difference.] Temp xy chrom coords Monitor RGB 3000 0.4373 0.4042 212 106 31 4000 0.3808 0.3770 173 117 65 5000 0.3454 0.3518 151 122 96 6000 0.3223 0.3320 137 124 121 8000 0.2954 0.3050 122 126 159 10000 0.2808 0.2885 113 126 185 12000 0.2719 0.2778 109 126 203 15000 0.2638 0.2675 104 125 222 20000 0.2565 0.2578 100 124 241 The resulting colors look exaggerated, particularly at the high end, but this is mostly due to the fact that the eye's perception of color from a point source is very poor. A star is a much better point source than a pixel on a frame buffer. Looking at a "defocused" photograph of stars will show that these colors are actually out there! I have not determined a way to "correct" the colors for this psycho-physical effect (other than pure cheating). As for the brightness of the displayed pixels, most monitors have a gamma near 2.5, and will therefore do a nearly correct exponential magnitude->brightness mapping for you! Thus do a linear mapping of say, mag 6 through -1, onto 1->255. You could probably work from the spectral class as well, but as long as you have a B-V value the above method is probably easier. It's fun by the way to make an HR diagram from the Yale data. - Phil <phil@brl.arpa> seismo!brl-smoke!phil