StarTools 1.8.518 Beta 1 now available

[admin]

I finally tried NBA in 1.8 last night - wow, had no idea you added this much new complexity on top of SVD.

Yeah, it seems like I just added a module, but it actually impacts many other modules as well.
StarTools is now essentially processing 3 datastreams in parallel (luminance, chrominance and NB accents).

I notice there's no exposure time setting for the NB file. Is that because it's superfluous, won't be used for L

Exactly. It is not used for luminance at all (not for chrominance for that matter). It is its own third "virtual" dataset.

I can confirm though the elsewhere-mentioned disconnect on the color/luminance/narrowband button in Wipe. When a NB file exists, the button acts as a tri-toggle. The button shows what is "next up" if you press it, and the description below the image correctly identified what is being looked at. However, if there is no NB file (i.e. normal old style), the button works properly but says "narrowband" when it should say "luminance." The below-picture description is correct though - luminance/chrominance.

Bugs. Bugs everywhere.

And this might better fit in feature request, but has it ever been considered to have the x and y slider end buttons run continuous on a hold, rather than single click only? Sometimes I run into this if I have ST in a window, such as to look at a log file on the side. The central slider area is pretty compressed, and so clicking on it can only get you into the general desired area. Then you have to use the plus/minus clickers to dial it in, and of course it's...click click click click...sometimes very many.

You can actually hover over a slider and use the scrollwheel on your mouse to quickly change values. Give that a try.

[Melty]

Hi Ivo,

Yes the dataset has been Background Extracted, Green Noise subtracted and Photometrically Colour Balanced in Siril. Probably in that order, if memory serves.

Under normal circumstances, I know that is a big no-no for Startools, but let me explain.

(In my hands at least) Startool's Wipe module doesn't do all that great with removing the multiple vignetting and colour gradients of the original data, regardless with the settings I care to use (I know, my data should be as untouched as possible before importing to ST). A full calibration workflow of flats, offsets and matching darks plus dithering at acquisition was also implemented, but I was using a fast lens at F/2.

I don't know if it's of any relevance, but it is also a multisession integration from various locations (although the same equipment and image scale), so I don't know if this is likely to confuse ST's tracking at all?

Thanks for investing this.

[Mike in Rancho]

Bugs. Bugs everywhere.

You can actually hover over a slider and use the scrollwheel on your mouse to quickly change values. Give that a try.

Aha. Well that will be useful and save some time. May also explain some funny things I've had happen, perhaps inadvertently scrolling or hitting a button whilst hovering.

Just little bugs. 1.8 alpha is quite well put together and useful. Particularly compared to all the 1.7 betas I couldn't get to work.

[admin]

Yes the dataset has been Background Extracted, Green Noise subtracted and Photometrically Colour Balanced in Siril. Probably in that order, if memory serves.

Indeed all rather big no-no's.

I have never come across a dataset Wipe couldn't tackle, though things may indeed become a lot harder in (rare-ish) cases where gradients undulate faster than the actual detail contained in the image (see docs). In those cases, subjective manual intervention (as offered by sample-setting capabilities) is indeed needed as algorithms (nor humans) can discern between detail and "gradient". Please see the docs on how to do this in StarTools (e.g. you work in the opposite way - you revoke samples where you are absolutely sure there is no background).

If true celestial gradients are severe, stacking multiple nights or stacking long sessions may see the gradients move and start forming complex patterns. This is pretty rare though, and most often the issue is truly with the flats - for starters, there should be no vignetting visible if your flats are working correctly...

If you absolutely must use some sort of photometry solution to balance your colours, simply take note of the balancing factors your application of choice comes up with, and then apply those factors in the StarTools' Color module. There is no need to compromise your dataset so severely from the get-go!

[Carles]

I don't know if it's of any relevance, but it is also a multisession integration from various locations (although the same equipment and image scale), so I don't know if this is likely to confuse ST's tracking at all?

Thanks for investing this.

Hi,

are you sure the focus was on point in both? did you change any setting on your camera? (gain or offset or ISO)
that artifact resembles one I had once when I stacked images from two different nights, but also different Gains.

if that's not your case...then I don't know. but you could just try to process it a big in Siril itself and see if the same
artifact appears (so image problem) or it doesn't (startools problem...? )

but as Ivo mentioned...those are all big no-no..

Regards.

Carles.

[Melty]

Hi,

This proved to be very difficult lens to generate flats for, even with my purpose made light box. Vignetting was sadly unavoidable to some degree.

With regards gain values, I used ISO 200 and 400, so that might explain that. I normally prefer to image at 400, but in this isolated instance due to the speed of the lens and the sky background brightness, 400 would have yielded too many super short subs, where more sky time would be spent dithering over actual imaging.

The last paragraph in the Wipe documentation points out that if the sky gradients undulate too much, then it may give problems. I feel that is what we have here, as if the output stack is stretched, there is literally a rainbow of sky gradients throughout the image. Hence my use of Background Extraction in Siril, beforehand.

I could upload the unmolested stack, if you require?

Cheers

[Carles]

Hi,

This proved to be very difficult lens to generate flats for, even with my purpose made light box. Vignetting was sadly unavoidable to some degree.

With regards gain values, I used ISO 200 and 400, so that might explain that. I normally prefer to image at 400, but in this isolated instance due to the speed of the lens and the sky background brightness, 400 would have yielded too many super short subs, where more sky time would be spent dithering over actual imaging.

The last paragraph in the Wipe documentation points out that if the sky gradients undulate too much, then it may give problems. I feel that is what we have here, as if the output stack is stretched, there is literally a rainbow of sky gradients throughout the image. Hence my use of Background Extraction in Siril, beforehand.

I could upload the unmolested stack, if you require?

Cheers

Hi,

the difference in ISO could explain the star artifacts... and the rainbow you mention, didn't go away after increasing Dark anomaly filter?

I also use Siril to stack and recently started using the script that stacks with background substraction, and StarTools seems fine with it.
I think, the other extra pre-processes you've used could be potencial problems.. Like white balancing and Green extraction. Since, for my understanding, Wipe expects a file without White balance and with some green noise, it might not know what to do with it ..probably simplistic way of thinking it, sorry Ivo

you can upload the stack if you like, if possible just stacked, or only background extraction done.


This how it looked when I stacked different gain files

Artifacts Gain.jpg (204.72 KiB) Viewed 5821 times

Regards

Carles.

[admin]

Hi,

This proved to be very difficult lens to generate flats for, even with my purpose made light box. Vignetting was sadly unavoidable to some degree.

With regards gain values, I used ISO 200 and 400, so that might explain that. I normally prefer to image at 400, but in this isolated instance due to the speed of the lens and the sky background brightness, 400 would have yielded too many super short subs, where more sky time would be spent dithering over actual imaging.

The last paragraph in the Wipe documentation points out that if the sky gradients undulate too much, then it may give problems. I feel that is what we have here, as if the output stack is stretched, there is literally a rainbow of sky gradients throughout the image. Hence my use of Background Extraction in Siril, beforehand.

I could upload the unmolested stack, if you require?

Cheers

Having the stack would be extremely helpful - I am always looking for problem datasets to base new development off. Thank you!

[admin]

I could upload the unmolested stack, if you require?

Many thanks again for sharing that stack with me! I'm rather unsuccessful, however, reproducing any issues with Wipe.

The Uncalibrated1 (or 2) presets do a decent job wholly unaided with, roughly, comparable result to the Siril stack (which would have introduced subjective signal discrepancies, e.g. not based on the data itself). It's probably worth experimenting with the Aggressiveness (reduce), Gradient Falloff (reduce), Synthetic Bias Edge Area (reduce) though, to better dial in the separation of gradient and signal.

I can see why the PCC in Siril had trouble - the chromatic aberration from the lens makes it hard/impossible to use the stars to calibrate against. This could also be the reason why Siril had trouble aligning the stars properly.

The Color module in StarTools likewise struggles with the initial calibration due to the aberrant color data in the highlights (you're not mixing ISOs by any chance are you? I'm seeing some other artefacts as well that are hard to explain, and are similar to the ones that tripped up the Contrast module...).

It requires some tweaking to bring out the brown dust and suppress the aberrant coloring in the highlights;

StarTools_2780.jpg (373.79 KiB) Viewed 5787 times

Please, please, please don't destroy your coloring/signal so early on in Siril!

[Melty]

That's a great result from the data, Ivo.

Prior to this, I had one last attempt using the unedited stack, and sure enough got a better result. I didn't make a note of my settings, but used my usual workflow. The Contrast module then worked as expected, so the additional Background Extraction, Photometric Colour Calibration and Green Noise removal in Siril clearly discombobulated Startools.

I couldn't quite bring out the same degree of dark nebulousity as the "Siriled" stack but it wasn't far off, tbh. Plus all the modules worked great so that's a plus. Not finished, as I am looking to add mono Luminance at some point.

Thanks for the tips, all.