1.51
#...
·
 |
|---|
|
I've been working on an RGB image of the Trifid Nebula, and I've been having problems with SPCC. The image solves just fine with ImageSolver, but the colors of the calibration don't look right. For example, there is too much red in the background. And the SPCC curves looked odd. Plus other issues. Today I watched an interesting YouTube video on another topic (https://www.youtube.com/watch?v=bw0WtNx0I8g&t=2708s), and the speaker mentioned that he sometimes uses SPCC after noise reduction and even after BlurX if using it earlier does not give satisfactory results. So I gave that a try. Normally, I like to use SPCC early... before other processes. Running NoiseX and BlurX did seem to improve the curves, and the color calibration was much improved. See the attached curves. Normally I run Blurx and then NoiseX, but today I did the reverse. No significance is implied by that change. So what do you look for when reviewing these curves? How do you evaluate the SPCC color calibration?  |
22.40
#...
·
·
8
likes
|
|---|
|
George, You are not alone in your struggles with SPCC. I've found a lot of variability with it. Sometimes it produces a result that "seems" spot on and other times it is clearly way off. I personally have a few things that seem to increase the odds of a good result: 1) Run SPCC before BXT or any other sharpening algorithm. Sharpening may change the color balance between the channels so you have the best chance of correctly matching the catalog data accurately when you use unsharpened, un-smoothed "raw" data. 2) It is important to make sure that the stars that are being used for color balance have a strong signal but none of the channels should be saturated (or over 90% of the range). Similarly, if the code uses very faint data or worse, warm pixels, that can throw the calibration way off. I've found that chaining the minimum structure size and the wavelet layers can play a big role in producing different results. So it may take some experimenting to find the correct settings for your images. I also believe that the quality of the subs that go into your integrated result may also affect SPPC. Small variations in brightness over the field due to passing high clouds can affect the result. Weeding out those images or using local normalization may produce a better result. I wish that I could tell you what works well each and every time but I haven't found the magic combination of parameters myself. So far, I just have to experiment with each image to see what I get. And no matter how it turns out, my images almost always show more green than I'd like, which often requires some compensation that screws up the fidelity of the final result. John |
