From: akelly5190 Sent: Tuesday, June 18, 2002 9:09 AM To: SBIG@yahoogroups.com Subject: [SBIG] Re: New LRGB procedure: Color Ratios LRGB Hi Paul, Bert, et.al., Please do not take my comments as a diatribe on your L/RGB processes or the fine processes so diligently developed by Rob Gendler and others, but I must finally make my statement about PhotoShop (PS) processing of astro-images, particularly color composition and luminance layering: while I think it is admirable that procedures have been developed (starting with Robert Dalby and Kunihiko Okano several years ago) to allow PS owners to use their software to process calibrated CCD images, it is well to keep in mind that this very complex, high-level, graphics-oriented software package was not designed specifically for high-S/N processing of low-flux color data. Its algorithms are not user definable and assignable at a basic functional level. While different concatenations of them (and folks are coming up with new ones all the time ) yield new results, *what is really happening to the image data at the pixel, ADU, and histogram level is usually unknowable*. This is particularly telling when processing RGB data, since the transition of RGB to HSL color space can be done in various mathematical ways and using the slider bars in PS creates HSL relationships which apply PS-proprietary algorithms, sometimes uncontrollably desaturating color by unnecessarily changing HSL relationships when layering. A few years ago Okano and Dalby published (on the Internet) their separate luminance-layering techniques developed on PS, starting the whole L/RGB trend. A very few imagers (including me) picked up on this immediately and tried it in PS, becoming quickly excited by the results and popularizing the technique. Among other things, I contacted Richard Berry, informing him of the advent, and complaining that while PS could do luminance layering, its functions were not very user definable and its rotation, registration, and other image manipulation algorithms were not up to those I was used to in Richard's Cookbook processing software and other true astronomical image processing packages, such as MaxIm DL, MIRA, and Megafix. Richard (and, soon, the other astro software writers) very quickly went to work to produce luminance layering capabilities in their packages, each doing it according to their own application of color- space conversion mathematics, but each (IMO) doing it better and in a much more user-definable way than PS. Now Richard and Jim Burnell have pretty much perfected (IMO) luminance-layering functionality for astro-images in AIP4WIN. I was a primary beta-tester for the AIP4WIN package, concentrating on basic image processing algorithms and the color processing modules, so I could reasonably be accused of bias; but I am not "shilling" for Richard and Jim (or anyone else) at all when I feel compelled to point out that those who want to do luminance layering the right way, with simple user-definable concatenation of RGB-derived H&S with a separately processed L and *without unacceptable loss of color saturation*, should obtain and use a true astro-image calibration and processing software. The cost of AIP4WIN (and most of the other packages) is so small compared to PS that I would be doing a disservice to recommend anything else. L/RGB processing is the way to go for astro-images. For those who think loss of color saturation is an issue, please see http://www.ghg.net/akelly/6563larg.jpg and http://www.ghg.net/akelly/gd27jhkl.jpg . These are recent L/RGB images that I have produced with AIP4WIN and posted to my website at http://www.ghg.net/akelly . The first (NGC6563 in Sag) was made last saturday night from unfiltered and RGB-filtered images taken with my SX MX916, Schuler RGBs, and 32" f4 Newt. The second (pre-main- sequence infrared nebula GGD27 in Sag) was made from JHK-filtered images (1.2, 1.6, and 2.2 microns, respectively) taken by Arne Henden of the USNO last month. In this image, the luminance was made from a stack of the JHK-filtered images. Finally, please do not think that I don't use PS and other graphics packages where they are useful (and they certainly are)! The images I post have almost always been exported as TIFFs from AIP4WIN, then pulled into PS for modest final "clean up" and conversion to publicly accessible JPEGs. By "clean up" I mean only 2 or 3 things: delicate application of a sharpening or faded minimum filter (IMO the single best function in PS), sometimes a subtle touch up of image brightness/contrast with the "curves", and final image crop and save. None of these things significantly change the basic luminance/RGB relationships and by-the-numbers color balance out of AIP4WIN. PS, Paint Shop Pro, and other graphics packages are great, but they need to be kept in their place..:). Kind Regards to All, Al Kelly