Help me improve this image

[Nyx]

Hi all. I am a fairly new StarTools user and could use some help.

The following is the calibrated/stacked result of about 1 hour worth of data, captured with a ASI533MC and a Samyang 135mm at f/2:

https://drive.google.com/file/d/1SrqKqf ... sp=sharing

And this is the best I can get using StarTools:

https://imgur.com/a/RaoD5kf

My StarTools routine is very standard: AutoDev > Bin > Crop > Wipe > AutoDev > Contrast > HDR > Sharp > Deconv > Life > Entropy > Tracking OFF > Repair > Shrink

Mostly using defaults. I invite you to process my image and put mine to shame. Any tips/help is appreciated.

For the record, the images were captured under Bortle 4 skies with very little clouds. I used AstroPixelProcessor to stack my data.

[admin]

Hi,

Would you be able to upload a single light frame for me as well?
I have some concerns about the stacks I've been seeing from APP lately, particularly with regards to color distortions, channel misalignments, normalization and pre-color balancing.

Is there any way to turn the white balancing and normalization off during stacking?

[Nyx]

Hi,

Would you be able to upload a single light frame for me as well?
I have some concerns about the stacks I've been seeing from APP lately, particularly with regards to color distortions, channel misalignments, normalization and pre-color balancing.

Is there any way to turn the white balancing and normalization off during stacking?

Hello Ivo,

here is a single light frame:

https://drive.google.com/file/d/1xoawx9 ... sp=sharing

I was extra careful when stacking with APP and disabled white balancing and background neutralization. I will also try stacking with DSS and let you know, if it gives better results.

[Nyx]

Here is the file stacked with DSS, using the suggested settings:

https://drive.google.com/file/d/1R0iCmT ... sp=sharing

And here is the StarTools result:

https://imgur.com/a/cYnfmss

The poor results are certainly due to the fact that I am still new to processing with StarTools. I am mostly going though the modules and selecting the defaults.

[admin]

The uploads are much appreciated!
It was very helpful; on closer inspection, the problem I'm seeing has nothing to do with APP or channel alignment. Rather, is the pretty severe chromatic aberration of the Samyang. Combined with some star deformity (see top/corners of the image), it can make it hard for stackers (and StarTools) to latch on to stars and detail.

The other thing, the yellow-ish color cast, which threw me off is visible the single frame too. It is likely due to not using an IR/UV filter (correct?), which sees the green channel response increase past the visual spectrum (see here).

If the chromatic aberration bothers you, acquiring narrowband is always an option too.

Honestly, it's not an easy object in the visual spectrum, and the chromatic aberration combined with this being a wide field makes it a lot trickier to process.

I ended up with something like this;

IC1396-NoSt.jpg (1.37 MiB) Viewed 6070 times

Processed in 1.7.412 alpha as follows;

Imported, using second option.
---
Type of Data: Linear and was Bayered, but not whitebalanced
--- Auto Develop
To see what we got.
--- Bin
To improve signal, use oversampling.
Parameter [Scale] set to [(scale/noise reduction 70.71%)/(200.00%)/(+1.00 bits)]
Image size is 1952 x 1959
--- Crop
Get rid of stacking artifacts.
Parameter [X1] set to [8 pixels]
Parameter [Y1] set to [12 pixels]
Parameter [X2] set to [1931 pixels (-21)]
Parameter [Y2] set to [1947 pixels (-12)]
Image size is 1923 x 1935
--- Wipe
Default.
--- Auto Develop
Default.
--- Wavelet Sharpen
Parameter [Structure Size] set to [Large]
--- SNR-aware Wavelet Sharpening
Auto-generate mask.
Parameter [Mask Fuzz] set to [10 pixels]
Parameter [Amount] set to [500 %]
--- Deconvolution
Auto-generate mask, though had to manually shrink it twice to avoid some artefacts .
Parameter [Primary Radius] set to [1.1 pixels]
--- Life
Default (reset mask to select all pixels)
Push back busy star field and re-focus attention on large scale nebulosity.
--- Color
Color constancy fails miserably here due to chromatic aberration.
The Legacy preset (Parameter [Style] set to [Artistic, Not Detail Aware]) mimics how less-sophisticated software such as PI, PS APP, etc. desaturates highlights as you process the image.
Parameter [Cap Green] set to [100 %] to remove any spurious green; chromatic aberration (incorrect focusing of different wavelenghts) tends to focus only green correctly, but not other channels, causing, for example star cores, to appear green, with blue and red energy scatterd across neighboring pixels.
Parameter [Highlight Repair] set to [4 pixels] to further repair star colors in over-exposing star cores.

It's just very challenging getting star colors to look ok, but at least we end up with a fairly reasonable representation of all black body radiation curve temperatures;

Selection_170.png (108.36 KiB) Viewed 6070 times

--- Wavelet De-Noise (siwtch off Tracking)
Parameter [Grain Size] set to [3.5 pixels]
--- Wavelet De-Noise
Default.

Hope this helps at all!

EDIT: You can also keep it simple, and not try to enhance the structures;

IC1396-NoSt-2.jpg (508.23 KiB) Viewed 6069 times

[Nyx]

Hey Ivo,

thanks for taking the time going through my files. You certainly got better results than I did.

I know that the Samyang gets recommended a lot for astro, but in my opinion focusing the lens and dealing with all its "imperfections" in post-processing makes it unattractive. I have processed (with Gimp) images shot using doublet APOs and they were a blast to work with in comparison to the Samyang. Anyway, kinda getting off-topic here.

Regarding filters, I am actually using an IR/UV cut filter from ZWO for my ASI533MC Pro.

The thing that "bothers" me the most when processing the image is actually the stars. I would love to see StarTools providing a way to separate stars and background, so that one can control how much each layer is getting stretched. I did search if there is a way to do something like that using layers in StarTools, but I still don't get it. It would be nice to have a way to semi-automate this when doing the 2nd global stretch. I am not sure if this is possible, but if it is, this is something to keep in mind for future versions of StarTools.

Anyway, I am waiting for a Zenithstar103 to arrive next week. Hopefully I'll have an easier time post-processing images taken with it than I had with the Samyang.

Thanks again for your help and thank you very much for creating StarTools!

[admin]

Hey Ivo,

thanks for taking the time going through my files. You certainly got better results than I did.

I know that the Samyang gets recommended a lot for astro, but in my opinion focusing the lens and dealing with all its "imperfections" in post-processing makes it unattractive. I have processed (with Gimp) images shot using doublet APOs and they were a blast to work with in comparison to the Samyang. Anyway, kinda getting off-topic here.

Stopping down can sometimes help ameliorate chromatic aberration. That said, if you have (or can borrow) a Ha filter, definitely have a go with that. Stars will appear a lot more controlled at your FOV, while chromatic aberration will - obviously - not be a problem.

Regarding filters, I am actually using an IR/UV cut filter from ZWO for my ASI533MC Pro.

Interesting... I cannot readily explain the very yellow initial bias then - I do not recall seeing this from other AIS533MC users, but it is indeed present in the single light frame as well.
Probably not something you should worry about, but more of a curiosity for me to investigate.

The thing that "bothers" me the most when processing the image is actually the stars. I would love to see StarTools providing a way to separate stars and background, so that one can control how much each layer is getting stretched. I did search if there is a way to do something like that using layers in StarTools, but I still don't get it. It would be nice to have a way to semi-automate this when doing the 2nd global stretch. I am not sure if this is possible, but if it is, this is something to keep in mind for future versions of StarTools.

I understand that this intuitively sounds like the way to go, and indeed StarTools has had this capability for years. There also other solutions (e.g. StarNet++) and procedures to combination of bothcombine their output with StarTools[/quote].

That said, great care must be taken, particularly in widefields like these that you don't end up with an image that mostly consists of "hallucinated"/inpainted data. Over the past year or so, I've seen people pass off StarNet++ images with stars layered back in as "photos" that look plausible at first glance, but on closer inspection contains vast amounts of detail that is not real around the areas where stars have been manipulated. It's a very unfortunate development and I personally hope the fad dies down soon.

Anyway, I am waiting for a Zenithstar103 to arrive next week. Hopefully I'll have an easier time post-processing images taken with it than I had with the Samyang.

Nice! Hoping for lots of clear skies for you!

[Nyx]

That said, great care must be taken, particularly in widefields like these that you don't end up with an image that mostly consists of "hallucinated"/inpainted data. Over the past year or so, I've seen people pass off StarNet++ images with stars layered back in as "photos" that look plausible at first glance, but on closer inspection contains vast amounts of detail that is not real around the areas where stars have been manipulated. It's a very unfortunate development and I personally hope the fad dies down soon.

I understand what you mean. I also find it ridiculous how many astro images are processed today to the point of being non-realistic. I think it's a fight you cannot win since almost any region of the night sky is bound to show a ridiculous amounts of stars that end up overshadowing the object of interest.

Nevertheless, it is IMO very important to be able to the find right balance between accentuating e.g. a nebula instead of the millions stars within the same FoV. This is why I said that it would be nice to have the option (in typical StarTools fashion, with sliders etc.) to make the stars less prominent my not overstretching them the same way you would do to pull out more detail for a galaxy or a nebula.

Usually, the Shrink module will suffice and it's definitely one of my favourite features in StarTools. Still, it is difficult to reduce star promicense AND keep it realistic on an image like the one I posted. It's just too many of them

Another interesting surprise was the Life module. Excuse my ignorance and lack of knowledge, since I still cannot wrap my head around much of the magic happening in StarTools, but using the Life module in that particular image seemed to tone down the stars in a pleasant way. I don't know how and why, but it was a pleasant surprise nonetheless. I have processed images of galaxies in the past where I skipped the Life module since, IMO, it did not improve the image overall. Again, it is very possible that I do not understand how to work with it!

[admin]

You're right! There are really roughly two ways that do not blatantly "make up" data, but rather manipulate what's there.
One is a morphological transformation (as used in Shrink), the other is multi-scale processing (aka [ur=https://en.wikipedia.org/wiki/Wavelet_transforml]wavelet transformation processing[/url]). Sharp and Life fall into the latter categories. The Life module's Isolate configuration can be thought of as a (very) distant cousin of the wavelet sharpener class of tools, working on extremely large structures, isolating them and manipulating them in isolation of the smaller sized detail. The documentation has an example of, like it did here, can push back a busy star field to re-focus the viewer's attention on the larger scale structures in the image;

[hixx]

Hi Nyx,
Do not give up on your Samyang lens: this is not typical behaviour. I own the same lens and found it to render stars tack sharp with no CA whatsoever even @ f/2. I also own a Zeiss Batis 85mm and found the Samyang 135 to be corrected superior for Astro. So this is not a bad lens.
Comparing the corners of your image, you can see slightly different star rendering, probably your lens/camera system is slightly deadjusted.
An option (if possible) is to use a tilting adapter between lens and the ZWO to adjust sensor plane vs. lens image plane.
But first you should try to stop down to f/2.8 This should improve image quality dramatically. Remember, no APO will ever perform @ f2 or f2.8, it is "stopped down" by nature hence much less prone to focus error, image plane deadjustment and some abberations. @2.8, your lens would still be great for large fields while super fast. The sweet spot for the Samayang by the way is between F2.8 and F4.
If all this does not help and your lens is still in the warranty period I´d recommend you to return your lens to the dealer for a new one. If not, you may want to have a service center adjust it.
clear skies