xonefs wrote: ↑Mon Oct 25, 2021 4:22 am
Hello, thanks for answering. It does and doesn't, since I'm still not sure if I can achieve what I want to in ST or how to go about doing it and where the limitation is. As far as scientific accuracy of color- well I'm not really concerned with that as most us are not scientists and choosing one of the complex percentage ratios of channels and adjusting bias to achieve a certain look doesn't seem much different than making some adjustments to taste. (how useful is a SHO 60% SII+40% HA, 30% SII+30%HA+40%OIII, 50%HA+50%OIII blend in conveying information or anything useful scientifically to a viewer?). I just want it within reason and to create separation. I'm not looking to invent pure fantasy but if the end result isn't aesthetically pleasing within a reasonable range, and on a level comparable to other images people are creating out there, then there's not much point in doing this for me.
I'm afraid you are missing the point; I'm trying to explain how/why the tools work the way they work so you can wield them effectively and to taste. It's not about being scientists or ideological esoterics.
The blends chosen are popular because they invariably choose colors that are opposites in the color wheel, maximising the usefulness of human color perception (for example blue and orange are perfect opposites).
Without understanding what the tools do (and why) you will just be pulling on sliders until you're "least unhappy". If you do understand what the tools do (and why) it is much easier to get to something you like
and that will be comparable to the work of others.
That looks totally plausible to me, and - colorwise - seems to approach Mike's earlier rendition quite closely.
I am equally referring to depth and detail and not just color. I see this detail in my channels before attempting to process in ST.
Now that you are (hopefully) across how StarTools processes luminance and detail are separately (one does not influence the other), we can now divide and conquer processing to your tastes much easier.
is that reasonable? Looking specifically at the structures around the edge look much better detail than after I processed or the other attempts I've seen here, and coloring seems natural.
That looks reasonable to me, and the detail is pretty easy to emulate.
To emulate that sort of detail, just pay attention to your dynamic range (global and local) and follow a workflow that progressively refines it.
Your dataset is actually much deeper than the examples you are giving.
To optimize the global dynamic range less for the extended detail, just specify a tighter RoI in AutoDev that excludes the extended detail (mostly Ha).
No RoI;
-
- StarTools_2792.jpg (265.96 KiB) Viewed 5583 times
Tighter RoI;
-
- StarTools_2791.jpg (256.07 KiB) Viewed 5583 times
Use Contrast for the address large scale local dynamic range discrepancies. For example, start off with the 'Local' preset and modify the Locality parameter to vary how local the dynamic range is optimized. Or use the default 'Basic' preset and, again, modify the Locality to taste.
-
- StarTools_2793.jpg (243.04 KiB) Viewed 5583 times
Next, run HDR. The Defaults ('Reveal' preset) tend to be a good starting point. I'm going with all defaults here as a baseline in most modules, but push detail as you see fit obviously.
-
- StarTools_2794.jpg (263.89 KiB) Viewed 5583 times
Run Sharp, then Decon;
-
- Selection_728.jpg (109.9 KiB) Viewed 5583 times
-
- Selection_727.jpg (153.95 KiB) Viewed 5583 times
Giving Decon the best possible signal you can muster is one of the reasons why you would want to bin to make use of the oversampling (which gives you a cleaner signal). As you can see this actually
gains you detail, rather than leaving it a blurry mess.
And then it's time for the Color module;
You can the usual blue/orange;
-
- StarTools_2795.jpg (369.69 KiB) Viewed 5583 times
Or a lesser known still SHO(!) palette;
-
- StarTools_2796.jpg (365.76 KiB) Viewed 5583 times
I like this palette, as it tends to yield coloring that is closer to visual spectrum while still being sort-of SHO.
Bigger version of the image;
Does that help?
It's clear I need the Contrast practice, and your data is great for it. Should work well on my Pacman also.
