Having done a few LRGB images now, I thought I was understanding the workflow and process. Some images my workflow seems to be fine, others the results are horrible. I've based my methods mostly on the 2 part LRGB video on the PI Youtube channel. In very rough summary:
- calibrate, register, integrate
- use LRGBCombination to just combine the RGB parts, leaving out the L
- BackgroundNeutralisation and ColorCalibration, DBE if needed
- any other linear processing needed
- auto STF both the L and RGB images
- drag the STF process icon onto HistogramTransformation bottom bar to transfer the STF settings to HT
- execute the HT and reset STF as shouldn't be needed now
- the histograms of all RGB channels and the L all peak roughly 1/4 point
- now combine the L into the RGB image with LRGBCombine
I've done this a few times, and on some images it works beautifully and simple. But other images the RGB by itself looks good, the L by itself looks good, but when they combine the result is horrible.
See below screenshot.
Bottom left is just RGB combined and stretched.
Top right is just L stretched.
Bottom right is the L extracted from the RGB image just to show how much brighter it is than the L integrated image.
Top left is the L applied to RGB using LRGBCombine. Horrible. I know I should play with the LRGBCombine transfer functions, but not matter what I adjusted them to, the resulting combo was horrible.
I'm thinking I'll just keep the bottom left RGB and toss the L. What do you think I'm doing wrong? Is it in my image capture or post-processing? All subs were same length and same binning.
Why is the L extracted from the RGB image so lacking in contrast compared to the actual L integrated image? To my eyes, the L top right looks more close to the bottom left RGB luminance.
