Processing IFN and dust : blotched images [Deep Sky] Processing techniques · patrice_so · ... · 7 · 567 · 5

Dear fellow astrophotographs, 

In my journey towards better astrophotography, I face many obstacles, making the journey thrilling. However, one obstacle is now becoming recurrent and I feel king of lost after having tried many ways to adress it. This issue is that my attempts at processing IFN and dust end up in blotched images. Here, a comparison speak better than word.

Both are crops of images on the left and the right were captured using an ASI294MC pro.  The left one is the one I admire. It was shot by @Konstantin Firsov and awarded IOTD 21.09.2023. See here : https://www.astrobin.com/gkfm6l/0/

The right one is mine. You will find the entire image here : https://www.astrobin.com/ssulcn/B/
This is unfortunately quite typical of what I end up with, even though I had some modest successes (https://www.astrobin.com/vjpfj9/B/). The issue is that sever that there are many image I did not dare putting online so far.
Focusing on the comparison here above, I can say the following about my image. Total exposure time reaches 29h45' at f4, which is arguably a good amount of light. Sky conditions were not bad. I am under a bortle 5 sky and nights were fairly clear. I removed about 10% of subs because of poorer quality. The entire processing was done in Pixinsight. WBPP stacking included drizzle and highest available quality. Calibration frames are prepared according to the books, including 5s dark flats as due with this camera. 

Processing wise, I processed the image as follow (letting stars aside here) 

And continued on a clone as follows 


Please note that I did not abused of LocalHistogramEqualization because i) each step was very modest and ii) the reversed mask had the effect that the process was applied to the main and brighter object, not to the dust at all. 

I tried many ways, but at this point, I would very much appreciate any help. 

Clear skies, 

Patrice 

just get more data; that way you can use less denoise and no overstretching, and you'll get less mottling

and yeah, keep the processing simple

also I recommend trying Arcsinh stretching a little bit; it brings out some color in dark nebula

My 2 cents.   Your mileage may vary.

Dump GHS from your process.  It sucks.   Take STF, relax the midpoint and blackpoint to very conservative levels where the noise disappears and the bright stuff isn't clipped.   Drop that into your histogram and apply to stretch your image.   From there, use Curves, very subtle steps each time.   Massage the data so the noise is minimal, but all your dust begins to pop.  Duplicate the image to create masks that can preserve key areas of your composition.   LHE is a great tool for an image like this, large scale functions and very low power. 


Every time you denoise, you're being destructive to your data.   Like it or not, theres no way around this.   I've changed up my workflows so that I do very little, if any denoise until the very end of the image.   Current denoise tools are extremely powerful, but like Uncle Ben said, "with great power comes great responsibility".    Be careful.   You may not notice the destruction right away, but you will as you progressively stretch the data more and more.    Don't go too crazy like I did on my 1333 image.  It's overly denoised…lol.   It's too late to fix it for me though.     Also, for what it's worth, even Konstantin's image is a little blotchy in the low signal areas.   It looks good from afar, but begins falls apart the closer you look.

It's definitely far less crunchy, particularly in the star forming region.   Much better details there!  Definitely an improvement.   I question how sharp the data is to begin with…. seeing a processed image says very little about the raw data.    I only wonder because I don't want to send you down a path that you can't achieve because the data doesn't quite hold up.

I also wanted to ask if you're properly color calibrating.   Particularly in the Ha regions of 1333, the detail feels very orange when there should be more red (Ha signal) to them.    Could also be that you're using ColorSaturation process and unevenly boosting saturation, which could throw off color calibration as well.