Please share your processing of a reference image

[prusling]

Hi,

Like many, I expect, I mostly use the standard module workflow with default parameters which generally serve very well. However there is always the thought that a tweak here or there, or perhaps the greater use of masks, could produce a 'better' result.

Could I suggest that those who are interested share the modules/parameters that they find work best on a given reference image. I propose that we use a preprocessed set of narrowband images of M16 that have been captured at First Light Optics' Ikarus Observatory in Spain. These have been shared by FLO at https://stargazerslounge.com/topic/3607 ... gust-2020/ on the following basis:

"We will be releasing all public data from the Ikarus Observatory project under a Creative Commons Attribution-NonCommercial licence. What this means is you are free to share, copy and redistribute the data as you see fit and adapt, remix, transform and build upon it but you must attribute 'Ikarus Observatory' as the source of the original data and must not use it for commercial purposes - i.e. no sticking images on the side of mugs and selling them."

I recognise that this is very high quality data and different techniques are necessary for dealing with more usual material. Also, that beauty is in the eye of the beholder, so one person's taste may be different from another's!

Please post your results and workflow/parameters. There is also a competition thread on SGL in the post linked to above to see who can get the most from this data and create the 'best' image.

To kick off, the next post will be my own standard result (stars could do with a bit more colour).

Cheers
Peter

[prusling]

Processed using 1.7.417alpha as follows (options/changes only - all others parameters at default values)
- Compose mapped as SHO to RGB (no luminance) with equal duration exposures
- Binned 50%
- Cropped by 20 px on all sides
- Autodev using a generous RoI around the target
- Wipe with 83% aggressiveness, 50% drop off point, 512x512 precision, 4 px dark anomaly filter
- Dev to 97%
- Contrast, HDR, Sharp with autogen mask (new feature) to 250% amount, Deconv with autogen mask with 1.9px radius
- Colour (mask filled) with SHO(HST) preset
- Track off / Denoise grain removal at 5px
Pillars could be made more prominent and star colours improved. Over to you!

Ikarus Observatory is the source of the original data - please see first post for conditions of use

[admin]

Not "best", but the minimum (using defaults and standard workflow) I would do to process this SHO dataset.

--- Compose
R = S-II, 14m
G = Ha, 15m
B = O-III, 9m
Note that the exposure times for this dataset are really in hours, but it's the ratios that are important.
--- AutoDev
To see what we got. Very clean data, no stacking artifacts.
--- Wipe
Narrowband preset.
--- AutoDev
Ignore Fine Detail < 1.5 to make AutoDev ignore some fine noise.
--- Contrast
Default
--- HDR
Default
--- Sharp
Default
--- Decon
Probably the one module where you cannot leave things to default values.
This dataset stands to gain a lot from tweaking (200% crop comparison);

StarTools_189decon.jpg (132.67 KiB) Viewed 6701 times

StarTools_187.png (506.2 KiB) Viewed 6701 times

I also used a conservative mask.
--- Color
SHO(HST) preset.
Removed some more green by increasing Green Bias Reduce to 9.00. This achieves that classic HST golden coloring.
Bumped up saturation a tad to 250%
--- Denoise (switch Tracking off)
Grain Size 7.0 pixels

Bonus tip; if you find purple stars offensive, launch the Layer module, choose "Invert foreground" as the Layer Mode, Keep, launch Color module, set Cap Green to 100%, then launch the Layer module, choose "Invert foreground" as the Layer Mode.

Full size result here

[prusling]

Many thanks, Ivo, I've worked through that process and achieved a similar result. The only difference I get is less noise reduction / smoothing at the end of the process - please can you advise any other selections/adjustments you make (e.g. is it grain removal or equalisation you use?)

[happy-kat]

I have been playing with the data as well. I wanted to know the camera used so I could select the optimised BIN rtio for the data, as I can't see what it is I did not bin it.
I have not used crop as did not see any artifices on the edge and neither did i wipe as it looked too clean.
I used compose but did a ratio of Oiii being 1 and ha being 1.35 and sii being 1.31 as that was what it worked out from the m,inutes and % of eachother assuming Oiii was the benchmark
My first process yielded magenta star bloat so planning to try again over the weekend.
I want to try to pull a little more contrast out on the snotty bits.
It is so different to anything I create to process myself.
I will be posting on the FLO topic when i get an image and the details

[Burly]

Heres my version mk1 ,i cant post log as st crashed not before saving , but pretty much followed ivos workflow few tweaks ,was going to use the flux module but caused a crash ,logged in bug posts
https://astrob.in/2zl4z6/0/

[prusling]

I understand that there is a script provided by PixInsight for 'dark structure enhancement' that has been demonstrated on this image. Is there a similar technique available in StarTools?

[happy-kat]

Only to say I have added my process to the SGL thread if you are interested.

[admin]

I understand that there is a script provided by PixInsight for 'dark structure enhancement' that has been demonstrated on this image. Is there a similar technique available in StarTools?

Apologies - only saw this just now. This would falls under dynamic range manipulations carried out by Contrast and HDR, however the Sharp module too has a DSO Dark preset which enhances only dark structures.

Cheers!

[happy-kat]

That works well, great tip thanks, though on my 3D render I prefer it less enhanced, but good to see it's effect on this target.