Weather has been nice and gave us two more clear nights here so that I could add some broadband data to the duo NB data of NGC 6914 which I collected some weeks ago. Blended via NBAccent this was the result:
NGC 6914 is actually the blue reflection nebula slightly right and up from the center of the image. It is embedded in nice hydrogen clouds and a colorful star field. Looks a bit like the Sadr region. I like that kind of stuff.
Deringing in SVD of 1.9.557 now worked quite fine. But my heart skipped a beat as ST just closed during the final denoise and all was gone...I tried to reproduce the error by processing along to the log...hasn't been reproducible. So hopefully it was my system that caused the crash.
Now this one actually looks quite good. Really nice stars everywhere, corner to corner. All the medium and larger stars are very round and seem to be nicely pinpointed with clean diffraction and spikes.
Setting aside NBAccent (if we can?), I wonder if there's anything we can learn as to why the datasets behaved so differently. With SVD acting on the synthetic luminance, I wonder if there are keys in the creation of that L. Here, I take it OSC or RGB is the base, with the other file being bicolor NB. Or maybe just cleaner SNR and better star shapes, thus better PSF outlines were created for sampling?
You're right, stars are looking better here from the get-go. But I don't know if there's a big difference in terms of deringing compared to the M 8 data set. almcl was able to dering the stars of M 8 just fine. So before we reason what's causing the difference we should be sure that there actually is one
I will have to try the M 8 data set again ...
Would you like to fiddle with the NGC 6914 data set(s)? So I would upload them as soon as I have them available.
Hey Stefan, well, I did try almcl's settings, at least to the extent provided, and didn't have much luck.
So, could be differences in the underlying data, the settings in SVD, or the processing leading up to SVD. I guess. Trying to digest all this plus the deconvolution explanations by Ivo recently to find out how to chart the best course for optimum SVD results.
I made a blinking comparison that maybe I can put in the other thread since that's the original source. But will also be curious how your own M8 reprocess goes.
But I have to admit I already thought the same ...