I came across this on the MaxImDL group. Really neat stuff. Go to the link on the first message and take a look at his curves and then read the second message on how he did it. Here are pieces of his original postings to MaxImDL: Hertzsprung-Russell diagrams... Posted by: "Robert Vanderbei" rvdb@princeton.edu vanderbei Sat Feb 6, 2010 7:32 am (PST) For the last few days I've spent some time constructing Hertzsprung-Russell diagrams from images of globular clusters (and some open clusters) that I've acquired over the years... http://www.princeton.edu/~rvdb/images/NJP/HRdiagMatlab.html I wrote a Matlab program (available from the link on the page above) to identify "all" of the stars in the image and compute their magnitude and color and then display an HR diagram. I wonder if I could have done some of this more simply and quickly in MaximDL. Any ideas? --Bob Robert J. Vanderbei Princeton University http://www.princeton.edu/~rvdb "Time's fun when you're having flies" -- K.T. Frog <http://groups.yahoo.com/group/MaxImDL/message/39931;_ylc=X3oDMTJydGQ0NXNuBF 9TAzk3MzU5NzE1BGdycElkAzMzOTkxMDMEZ3Jwc3BJZAMxNzA1MDgyODA2BG1zZ0lkAzM5OTMxBH NlYwNkbXNnBHNsawN2bXNnBHN0aW1lAzEyNjU1MzU1Mzk-> Re: Hertzsprung-Russell diagrams... Posted by: "Robert Vanderbei" <mailto:rvdb@princeton.edu?Subject=%20Re%3A%20Hertzsprung-Russell%20diagrams %2E%2E%2E> rvdb@princeton.edu <http://profiles.yahoo.com/vanderbei> vanderbei Sat Feb 6, 2010 8:22 am (PST) Hi Bill, The first thing I did was to recreate a pure unadulterated color, floating point fits file using my original stacked R, G, and B frames (in the past I didn't always save a color fits file and when I did I sometimes applied some sort of stretch or sharpening to it before saving). My Matlab program then reads in this color fits file. The Matlab program splits the color image into R, G, and B frames and makes an L frame as the average of the three color frames. It then looks at each pixel in the L frame and tries to determine if this is the central pixel of a star. To be a star pixel, the adu must be larger than at each of the 8 nearest neighbors and some of those neighbors are required to have larger adu values than their neighbors further removed. In other words, the pixel must look like a local max. Then I compute a few different estimates of the overall flux, for example, by averaging the peak pixel and its four nearest neighbors and subtracting an average background level computed as the average of the neighbors of the neighbors. I drop from my star list any candidate whose flux is too small (to prune out random noise). This gives me my star list. Then, I compute color as the logarithm of the ratio of the R value to the B value at the peak pixel. From the flux and the color, I make the g raph. Hope this is the sort of detail you were looking for. --Bob Jerry Foote ScopeCraft, Inc. 4175 E. Red Cliffs Dr. Kanab, UT 84741 435-899-1255 jfoote@scopecraft.com