Page 3 of 3

Re: M27 - need a hand, please!

Posted: Sun Oct 14, 2018 1:12 am
by devonshire
Richard,

I updated to the latest version.

That really was a very helpful video - I'll be referring back to that. Attached is a jpg of where I ended up after trying to replicate what you've done (first full run-through).

There were a couple of things you wondered about during the video...
The noise is from my backyard sky - I think I'm in an Orange zone.
The image is a dss stack (save as, don't apply), from my Nikon D5300 DSLR, so that explains the size.
The green in some of the stars may be from my scope. It's an old (2002) Stellarvue 102EDT. A really good scope for its time, but not an APO, and I think that mostly shows up in both the size, and internal color of the stars.

If I keep at it, I think most of what you've done will sink in eventually :-), but the one step that's just a complete surprise is the effect of ROI in Autodev. The star you selected does indeed clean things up, but selecting others produces different effects, and of course, that star isn't really the point of interest (M27). I looked back at the doc, but no lights went on. Best guess is that it guided the stretch for the star field, and M27's appearance was just a fortunate byproduct. What do you think?

Thanks again for all your help!

- Bob

Re: M27 - need a hand, please!

Posted: Sun Oct 14, 2018 3:52 am
by admin
Hi Richard,

Further to your PM re: video, the link is working now. Thank you so much for helping out a fellow ST user with a great video!

A couple of notes;

You were wondering exactly what HDR is doing, specifically the Reveal algorithm.
The revamped Reveal algorithm is a fairly new addition to ST.

It creates a local histogram equalize. The "local" part means that it doesn't do this for the whole image, rather it does this for a sliding window. You can see what this looks like in the Layer module, when you select Local Histogram Equalize. Next, it compares that "image" to the input and calculates how "far-off" the input image is per-pixel in terms of contrast vs the immediate neighborhood. It then uses this measure of contrast to "nudge" the source image to the local histogram equalized version. It further takes into account signal-to-noise constraints as informed by the Tracking data. The algorithm is loosely related to Contrast Limited Adaptive Histogram Equalization algorithm, except that the contrast limiting and noise propagation part is far more sophisticated and optimized specifically for non-undersampled detail as found in atmosphere and diffraction-limited nebulosity and galaxy datasets.

The Contrast module is best used on large objects or screen-filling nebulosity; it is best used for medium-to-large scale dynamic range optimization and manipulation.
As a rule of thumb, going from global>large/medium>small scale detail enhancement, use AutoDev/Develop>Contrast/Life>Sharp/HDR respectively.

Color/sampling of stars; if you calibrate on a foreground star field, make sure you include the stars' halos. Green is indeed rare in outer space, however make sure you don't pull back on the green channel too much. Green noise - just like red or blue noise - is natural once color calibrated. Getting rid of spurious green pixels is not the goal of color balancing. If you need to to get rid of spurious green pixels and the color balance is otherwise good, use the Cap Green parameter.

Thanks again!

Re: M27 - need a hand, please!

Posted: Mon Oct 15, 2018 12:34 am
by devonshire
Richard, Ivo...

Any thoughts on that ROI in Autodev?

<<
If I keep at it, I think most of what you've done will sink in eventually :-), but the one step that's just a complete surprise is the effect of ROI in Autodev. The star you selected does indeed clean things up, but selecting others produces different effects, and of course, that star isn't really the point of interest (M27). I looked back at the doc, but no lights went on. Best guess is that it guided the stretch for the star field, and M27's appearance was just a fortunate byproduct. What do you think?
>>

Re: M27 - need a hand, please!

Posted: Mon Oct 15, 2018 4:02 pm
by Rkonrad
Given what I recall about Ivo's initial response to your image - I think you need longer light frame exposures to get the best our out of your dso. My roi on the large star is just experimentation and largely a rescue operation which I have done for my own work - most of the time (no tracking and lack of time imaging). I tried roi on the main object but the stars and noise seem to dominate. When I increase the roi on the large star to maximum, it also reveals the smaller stars as well. I've tried other options but this seems to work for me.

Cheers

Richard

Re: M27 - need a hand, please!

Posted: Mon Oct 15, 2018 4:07 pm
by Rkonrad
Nice job Bob! Couldn't have done better myself.

Richard