StarTools

Further to my previous comments about HDR in version 1.8 , I decided to process some comparison images using both ST1.7 HDR and ST 1.8 HDR default settings only using the same data set ( IC 2944 Running Chicken Nebula)

Capture details are on my previous posted image of IC 2944 including my Astrobin link

Processing details for both ST1.7 and ST 1.8
Data set loaded via Compose ( OSC / DSLR Bi Color )
AutoDev for inspection only
Bin 65%
Crop
New Image size 3892 x 2565
Wipe NB, dark anomaly 2 pixels
AutoDev ROI , Gamma 95% , ignore fine detail 2.5 pixels
Contrast ( Darken )
Image 1 No HDR
Image 2 Reveal Default
Image 3 Optimise Default
Noise reduction 4 pixels
Saved as jpegs

No Sharpening
No SV Decon
No Shrink
No Color
Etc………..

From these comparisons the default settings of ST 1.7 HDR Reveal and Optimise work really well and only require minor tweaks if at all
ST 1.8 HDR Reveal and Optimise results in the main structures looking like shattered glass.I have to really reduce settings to absolute minimum to achieve something similar like ST 1.7 HDR but not as good as ST1.7 HDR
From my previous target data sets and in particular this image ST 1.7 HDR provides a better outcome
In regards to Tame preset in ST 1.7 and ST 1.8 HDR , both work well but I do prefer the results provided by ST1.7 especially HDR DSO Core

Comments welcome
Thanks

ST 1.8 HDR images below

All images down sized to under 500KB to enable upload

Clear Skies
Martin

Totally agree with you, for certain datasets and lead-in processing, and can't argue with your real-world examples as they show what they show. But what are you shooting for, that the default settings be altered?

There's just so much that goes into customizing HDR now, that one-size fits all may be less relevant than with prior ST versions. And it may also be rather important how one did Binning, OptiDev, and Contrast.

I would not object to a softer right-side Tame default, as in many cases I will drop it back. But of course it depends if one does a lot of galaxies, or wide field nebula, or clusters...

I was thinking maybe the new Compose image identification options could perhaps feed differing defaults into HDR, but we'd probably have to do a lot of testing as to each such image type to come up with some reasonable sets of settings. We could all post proposed defaults and then vote! Unsure how that would work together with HDR already having its own preset buttons. An array?

Not all wide-field nebula are the same though. I just loaded up my questionable year-old M42 stack for scientific purposes. And because Ron's post made me realize it's not looking good out my way for a 2023 acquisition.

This is APS-C size data, binned 50%. Wipe, OptiDev, Contrast. Then in HDR I just left everything alone - 50x50, 30, 30, Tame 1.25. Not only no shattered glass at all, but I'm probably going to have to bump Tame to 1.50 or more to hone in on the Trapezium a bit more, and might up one or more of the boosts as well.

I can confirm Martin's observations. I had an easier time using 1.7 HDR. And I sometimes see the 'shattered glass' effect with the defaults in 1.8/1.9 like Martin described. But I always use HDR nonetheless. Fiddling with the sliders almost always leads to a valid result for me.

Regards
Stefan

Hi Martin,

I'm having a look at this now, and I am trying to understand how to replicate this. I am wondering if it is perhaps caused by the scale you are processing at, and whether the Context Size parameter makes a difference here?

Hi Ivo,

regarding this, please have a look at this discussion as well:

viewtopic.php?f=4&t=2736&sid=fd80c131f5 ... 6cb2d309e3

Best regards, Dietmar.

admin wrote: ↑Tue Mar 28, 2023 4:44 am Hi Martin,

I'm having a look at this now, and I am trying to understand how to replicate this. I am wondering if it is perhaps caused by the scale you are processing at, and whether the Context Size parameter makes a difference here?

Hi Ivo,
My nominated scale size is usually based on a scale where SV Decon works best for my images ( so a balance between Binning and Cropping )
I guess the main focus of the comparison was generally using “default settings” in both HDR 1.7 and 1.8. Im sure if I reduce the Context size in 1.8 the outcomes would be different.
Thanks
Martin