Hello,
I would like to share here my winter's project, M42, finally finished.
I posted this image in another forum, so I do here a copy-paste (more or less).
I followed a workflow that I found on that forum with minor variation, I would be glad if the experts could give me feedback about it.
The image:
Link to Astrobin:
https://www.astrobin.com/398087/0/?real=&nc=userEquipment:
Borg 101ED f/4 + QHY163M + Optolong LHRGB + IDAS LPS-P2 + HEQ5
Adquisition:
L: 10h (613x60s+50x5s+50x1s)
H: 21h (214x300s+50x120s+98x60s+82x10s+45x5s+49x1s)
R: 2h (132x60s+48x5s+48x1s)
G: 2h (133x60s+50x5s+50x1s)
B: 2h (118x60s+49x5s+50x1s)
Processing:
The stars suffers from a mix of halos, microlens artifacts, chromatic aberration, noticeable in the blue channel mainly, and astigmatism, bad cocktail to deal with.
In addittion, I wanted to experiment with short exposures but I think that I went too far... the RGB data is not as deep as I expected and I suspect that the background didn't overwhelmed the read noise.
Processing Workflow (asuming calibrated, registered and cropped L,H and RGB MasterLights):
RGB (Linear)
1. PCC to the RGB image
2. PixelMath to combine H with RGB with Max operator
I prefer the result of this process over NbRGB Script.
HRGB (Linear)
3. MLT Sharpening (RGB Splitted / Masked)
Applied to the individual RGB channels, this is very important, in my opinion. Every channel has his own SNR, so I consider mandatory to use a specific mask for each channel.
I apply bias only to the 2 first layers as I only want to enhace very fine details.
3.1. DBE with fixed samples
This step is not necessary if the image is free from gradients, but is not my case.
The difficulty for applying a DBE with this images plenty of nebulosity is that there is no room for placing samples. I was investigating more deeply how the different paramenters of DBE works and realized that the values of the samples can be fixed. This method opens a new world of possibilities and gives a highly accurate control over the gradients.
I seached for information on this way to use DBE without success. If someone has more info I would be grateful to know more about this. Anyway, I will continue investigating and I will probably write a tutorial.
4. HT to clip highlights
After PCC the cores of my stars became reddish. I saw that the RGB values inside where not 1,1,1. They where not clipped. That makes almost impossible to saturate them without resulting in artifacts.
To clip higlights is bad, I know, but in my opinion the benefits outweigh the disadvantages, so I consider it as a very important step.
In this image I did this process after MLT Sharpening, but maybe its natural position is just after HRGB combination. I have to investigate it.
5. MLT Noise Reduction (2 Steps / RGB Splitted / Masked)
Again a process that I consider that has to be applied to the individual channels for the same reason as above.
I make this NR in two steps, first one for the blackest pixels, second one, the usual application.
6. Saturation
Typical saturation enhacement before NLT.
7. Standard STF to HT (Non Linear Tranform)
No MaskedStretch, ArcSihn... Specially in a image like that, with this huge dynamic range.
In general I prefer to deal with a bloated, but naturally stretched image, but I understand that is a matter of taste (or workflow preference)
HRGB (Non Linear)
8. SCNR
Typical values Green, Average Neutral, 1.00. Worked well in this data.
9. Curves to restore the core (Masked)
I spent over a week trying to apply HDRMT but always the same, colors completely washed, so I went to the old but effective workflow of itteratively mask and apply curves to restore the core I used 4 steps in this case.
In my opinion HDRWT works fine for L, but not for RGB.
10. Blue Halos Correction (RGB Splitted / Masked)
Specific process to my non-perfect optics. I splitted the RGB channels, and applied a erosion to the halos with a specific mask. Worked well, but if some one knows any good workflow it would be great to ear about it.
11. Pre-LRGB Combination
I had to make minor adjustments to the HRGB data before LRGB combining. This includes, a HT, DBE and Curves.
L (Linear)
1. MLT Sharpening (Masked)
Same parameters as for HRGB
2. MLT Noise Reduction (Masked)
Performed this time in 1 step (the data was very good in terms of noise).
3. Standard STF to HT (Non Linear Tranform)
Same workflow as for HRGB
L (Non Linear)
4. HDRMT to restore the core
This time yes, HDRMT worked extremely well here. Applied 6 layers with lighness mask checked.
5. Local Histogram Equalization
I felt in love with this tool since the first time I used. It's amazing!!
I like so much to enhace the background. In this case I used it to the whole image with a very big Kernel radius.
6. Star Erosion
To reduce the size of the stars is not my strong point, but this time I improved my workflow by using a mask on wich I excluded the cores from the erosion. This helped me to avoid the typical "dead stars" I don't like them at all.
7. Pre-LRGB Combination
It consisted in to apply a DBE.
LRGB (Non Linear)
1. LRGB Combination
Applied with standard parameters.
2. ACDNR (Chrominance)
Applyied in a very small amount, with very conservative parameters to keep as detail as possible from the RGB channels. In my opinion this is key to achieve a good representation of the different structures.
3. Final Tunning
Some processes to get the final result.
-DBE to remove remain gradients. I used again with fixed values in an iterative process.
-Saturation
-Final Application of a Standard STF to Histogram Transform.
C&C are welcome!
Thanks for watching!
Alberto.
