New starnet alternative and use with startools workflow

General discussion about StarTools.
Post Reply
xonefs
Posts: 64
Joined: Sun Jul 18, 2021 7:30 am

New starnet alternative and use with startools workflow

Post by xonefs »

I played around with the new starxterminator for photoshop and it seems to have impressive results- much better job than starnet with the artifacts.

I wish there was a way to integrate it into a startools workflow but I don't know if there is any proper way to do it. It would be nice to process starless data since often even with star masks my stars sometimes just get slightly weird. I saw some old posts about starnet and startools on linear but it sounded like a huge pain and especially for me on a mac. this starxterminator plugin works well and is super fast.

Is there any way to export individual channel files from startools after wipe and some other tools like decon? I know there seemed to be some resistance to offering this. Would I have to save it as an RGB tif and then open it up in photoshop and extract individual color channels? would that mean I would have to color it first?

It would be quite nice if I could export after a certain point and bring back in. It would be nice to work all in ST but I just don't think I can yet and working on starless data would be better.
User avatar
admin
Site Admin
Posts: 3367
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: New starnet alternative and use with startools workflow

Post by admin »

Hi,

There is no place in a StarTools workflow for star removal, because star removal is a "solution" to a problem that doesn't exist!

While it may make intuitive sense at first, it does not - at all - yield anything valid from a mathematical, nor signal processing, nor photographical point of view.

The reason why such "separation" (between quotes because it is not truly!) manipulations always immediately stand out in an image, is because everything in your image should be affected by the same Point Spread Function ("blur function"). In separately manipulated images, this is no longer the case. (It's also worth noting that a PSF for a single pixels affects the entire image!).

Disconnecting some parts of the image from other parts of the image, will cause the two "separated" images diverge wildly in the way the PSF is retained and shaped. Worse, some operations are performed based on inpainted data, the processing of which will progressively bleed into neighbouring pixels.

Star removal is the next best thing if your software cannot be made star-aware by some sort of masking capability (indeed, star removal software definitely has legit uses for creating star masks and is chiefly used as such in apps like PI, with PS now also having a useful tool for this). It has absolutely zero value beyond that for serious processing. All tools in StarTools are, however, star-aware, and you should not need an external tool (do let me know though if you find the opposite of course - I'd be more than happy to look into why your stars get "slightly weird"!).

The resistance to the suggestion of incorporating this sort of procedure into a workflow, is not a philosophical one; it is purely a practical one. It is an arbitrary manipulation (much like selective processing with a mask) that has no basis (e.g. cannot be modelled, explained or expressed) in physics, mathematics or signal processing. As such, it is impossible for something like signal evolution Tracking to make any sense of signal that materialises out of nowhere; selective manipulation never had a place in astrophotography (astro art not withstanding), simply because it does not constitute a transformation of your signal. Procedures like these invariably add signal.

It is a free world of course, but I implore you to leave selective processing by the way side if you wish to progress as a photographer.

Hope this helps!
Ivo Jager
StarTools creator and astronomy enthusiast
xonefs
Posts: 64
Joined: Sun Jul 18, 2021 7:30 am

Re: New starnet alternative and use with startools workflow

Post by xonefs »

Maybe it's the implementation of the star masks then, since I do find sometimes they do not work to fully protect the stars from getting larger after processes. I also find use of the color module in ST challenging in regards to saturating narrowband data without oversaturating stars, or trying to process HOO data and remove the color cast from stars in ST to have more white stars. I've tried using inverted star masks and i find the autogen starmask function does not do a good job of creating an even round star masks in the color module and they look oddly shaped and pixelated sometimes, or don't select enough stars.

I'm not married to the idea of star removal, but it seems star masks are also a workaround as much as removing stars and replacing them to achieve certain things. Maybe I just want better and easier star masks.

Slightly unrelated but I also wish it was possible to work on color after denoising, or be able to denoise and go back to tracking to work on color and have a denoised view- I find it challenging to process color and get a feel for how the final image will look when you are looking at a noisy image and trying to process color and seeing all the noise mixed in when I am trying to focus on color.
Post Reply