Blotchy data when stretched.

[CraigD1986]

Hi All.

Is anyone able to shed any light on why this image is coming up so blotchy? The attached screenshot shows the Screen Transfer Function in Pixinsight but I get similar results when I start gently stretching using GHS. The process so far is as follows:

Crop, DBE, BlurX, NoiseX, StarX.

There is about 8 hours of data from a ASI533MC Pro with Optolong L-enhance.

[Astroscot2]

Hello Craig.

Im pretty new to PI and using GHS,  I think possibly you are stretching the background too much, I think its easily done done in GHS. I dont have great success using it .

 

Mark

[CraigD1986]

The above is just a STF visual so the data is still linear. I get blotchy results even using EZ soft stretch

[vlaiv]

I'd blame things with X in their title

Try simply removing those last three entries from your workflow to see what you get?

Just do Crop and DBE to start with.

[Adam J]

34 minutes ago, vlaiv said:

I'd blame things with X in their title

Try simply removing those last three entries from your workflow to see what you get?

Just do Crop and DBE to start with.

Its more like miss use of things with X in their title. 

Adam

[CraigD1986]

Wow, thanks @vlaiv and @Adam J. I've gone back to pre-RC astro plugins and then ran Starnet 2, reset the STF and here is the result. It looks like one of the X's is causing the problem. I used the default settings on each of the plugins. My problem is now that I was using the free trial on all three plugins, which have now expired so I can't experiment further without purchasing all three. This experience doesn't exactly make me want to purchase then.

I've heard nothing but good things about Star Xterminator, Noise Xterminator and Blue Xterminator though so now I'm a little confused.

[CraigD1986]

Okay, now I'm getting confused. I concluded that the problem was with Noise Xterminator but I've also tried EZ Denoise and the same problem occurs. This screenshot shows the same image, processed in the same way (Just crop, DBE and Starnet 2) and a STF applied. The image on the left has had EZ Denoise applied and the image on the right hasn't. When I reset the STF after the noise reduction, the image looks horrible. I get similarly bad results if I use a basic stretch like EZ soft stretch.

[Adam J]

2 hours ago, CraigD1986 said:

Okay, now I'm getting confused. I concluded that the problem was with Noise Xterminator but I've also tried EZ Denoise and the same problem occurs. This screenshot shows the same image, processed in the same way (Just crop, DBE and Starnet 2) and a STF applied. The image on the left has had EZ Denoise applied and the image on the right hasn't. When I reset the STF after the noise reduction, the image looks horrible. I get similarly bad results if I use a basic stretch like EZ soft stretch.

So your main issue is that the black level is really far too high. I normally aim for a black level of around 25-30 on the 0-255 scale.

I dont use wither of the X's I use Topaz Denoise and only on its mildest settings as if you push any of these too high then its going to result in the image quickly starting to look atrificial and details being extracted from the noise. This looks to me like oversharpening.

I cant help on the setting you should be trying though as I simply dont use those programs.

Adam

 

[ollypenrice]

A lot of the noise is green and would be helped by SCNR green.

Once I've got the background up to the level I want it (with the black point brought in to the start of the data line) I don't stretch it (the background) any more.  What would be the point? I'd be dragging it above the nose floor. Instead I continue to stretch in Curves (now available in PI, I think?) with the background level, and below, held by fixing points.

I'll also continue to stretch through a mask made from the image itself, but with the contrasts exaggerated.

Essentially, I don't want to stretch the fainter stuff by any more than I have to. Stretching it and bringing in the black point, repeatedly, as seen on many tutorials, strikes me as being a way of boosting noise.

Olly

[Laurin Dave]

I've seen something similar when I've run StarX prior to stretching. My workflow is run BlurX, then stretch till the Stars are the size I want..  remove stars using StarX,  further stretching using similar technique to  @ollypenrice then colour manipulation, any further local contrast enhancement and/or sharpening, colour noise reduction, then NoiseX if necessary, then add the stars back in screen mode prior to a final polish.  Maybe try that and see if it helps.

Dave

[The Lazy Astronomer]

Bear in mind that AI star removal will very likely create artifacts. These should be largely (or completely) obscured once the stars are put back in - I often quickly add my stars back in (and undo to remove again before doing next thing) at various points throughout the process so I can see how the final, starry, image will look. 

[ollypenrice]

2 hours ago, Laurin Dave said:

I've seen something similar when I've run StarX prior to stretching. My workflow is run BlurX, then stretch till the Stars are the size I want..  remove stars using StarX,  further stretching using similar technique to  @ollypenrice then colour manipulation, any further local contrast enhancement and/or sharpening, colour noise reduction, then NoiseX if necessary, then add the stars back in screen mode prior to a final polish.  Maybe try that and see if it helps.

Dave

Ah yes, good point. I don't de-star when linear. I'd rather see any issues with the starless right away than find them only after the stretch. I like to do a roughly 2/3 stretch before de-starring so I can exploit the advantage of a starless final stretch.

Olly

[gorann]

Yes, it is possibly a result of doing StarXT on the un-stretched image. I never tried that but like Dave and Olly I stretch the image first so I get the stars at a level I like and then run StarXT. That may be the reason why I never run into any blotch similar to yours. I have more hard to believe that NoiseXT would be the problem - it has been quite subtle on my images (I use the standard settings).

[ollypenrice]

3 hours ago, gorann said:

Yes, it is possibly a result of doing StarXT on the un-stretched image. I never tried that but like Dave and Olly I stretch the image first so I get the stars at a level I like and then run StarXT. That may be the reason why I never run into any blotch similar to yours. I have more hard to believe that NoiseXT would be the problem - it has been quite subtle on my images (I use the standard settings).

Yes, I can't suspect NoiseXt, which is as close to perfect as anything ever gets. I run it as a bottom corection, top layer then go to the sharp, detailed parts of the image to erase it from there if it's done any damage - but it hardly does any at all. It's rarely worth erasing it from the detailed parts.

Olly

Edited February 25, 2023 by ollypenrice
Correction