Hi,
since I participated in the discussion
http://pixinsight.com/forum/index.php?topic=1074.msg5000#msg5000 that finally gave us the Color Calibration Tools in PI 1.5, I could not resist to give PI 1.5 a quick try tonight. What sparked my interest in color calibration was the difficulty to get a "true" color for the core of Comet Lulin.
So today, I quickly reprocessed my stacked image with DBE (using the immediate subtract available in PI1.5). I used the new Color Calibration Tool, and arrived quickly at a white balance that is almost as good at anything I got after hours of manual experimenting months ago (I am sure I can get the tool to be better with some more tweaking). I then stretched the histogram using 2 methods. I did not do any noise reduction etc. yet.
http://cid-c56e60845cfa6790.skydrive.live.com/self.aspx/Pixinsight/lulin%7C_pi15.jpgThe left picture is the usual non-linear stretch using the Histogram Transformation Tool. The right one uses the following procedure:
* Split image into HSV using Channel Extraction Tool
* Stretch V-Image using Histogram Transformation Tool
* recombine HSV to RGB using Channel Combination Tool
So this is a kind of artificial L-RGB process.
I am the first to admit that the images leave much to be desired, but I believe that
* I have better control over the red background noise that is visible in the left version
* the comet core and many stars (e.g. in Preview 3) are more colorful.
In fact, looking at the RGB values of the unprocessed image, it is clear that the mathematics of a non-linear histogram transform of an RGB image loose color in the bright parts, such as the comet core and some starts. The relative values of R:G:B get closer to 1:1:1. So this LRGB version may be helpful in such cases.
Georg
