Can't seem to reduce the noise without it getting "Blotchy"

[STAstro]

Hi All

I've spent quite a bit of time on PixInsight over the past year, but M81 / M82 is really causing me to loose what hair I have left on my head, I can't seem to minimise the background noise without it getting blotchy, this is what I tried so far

1. Performed a DBE on LUM Image
2. Auto STF on LUM image
3. Clone out to secondary image
4. Apply Auto STF to HT and apply to Clone
5. Use Clone as Inverted Mask on LUM image

Now I have tried ACDNR, MLT, MMT all with mixed results, attached is my Luminance Image, I have an RGB and HA image to process as well, but I am starting with the Luminance image and maybe I could learn where I am clearly going wrong

The Fits file can be downloaded from here

https://www.dropbox.com/s/tajcfqwpjzszmbn/L%20-%2060x300S%20-%20DBE.fit?dl=1

Looking to learn

Simon

[Greg Schwimer]

I think the blotchiness may actually be IFN.

https://apod.nasa.gov/apod/ap080512.html

[STAstro]

Thanks for the reply, it's not IFN, I can see some IFN in the image, but it's definitely not that

[Tromat]

Hi,

did you tried TGV, like the example I posted today here: https://pixinsight.com/forum/index.php?topic=11971.0 ?

[STAstro]

Thanks Tromat

Just trying that now, I presume you are applying the mask inverted right?

Simon

[msmythers]

Simon

A 50 percent gray mask does not need to be inverted. It's 50 percent of an all gray screen so it's the same mask inverted as not.


Mike

[STAstro]

TGVDenoise with the 50% Mask made no difference to the image whatsoever, here's the before image

[STAstro]

and here's the after TGVDenoise applied with 1000 iterations

[msmythers]

You can use TGVDenoise in a small preview and you can then fine tune the edge protections and other parameter in that preview. Make sure your mask is applied to the main image first. Also while fine tuning the edge protection start with a low number of iterations, say 25 and raise that once you find an edge protection value that seems good. You can use the Statistics tool and the stdDev value as the starting point for the edge protection. I drop the value by 2 decimal points to start. I also like to start around 2.5 for the Strength setting and a value of 2 for the Smoothness. All parameters can and most likely need tweaking if you want the best results.

Here is what I got with a very quick try with TGV and only 25 iterations.


Mike

[STAstro]

Hi Mike

Thanks for your help, now this is really weird, I can see the changes in your screenshots, but when I apply those changes to my "Preview" of the same area, I see no difference between the before and after, so the question here is what have you done differently?  I am going to start all over again with the mask!

Regards
Simon

[STAstro]

Ah F*!@

I just realised when I created the mask, I did a pixel math of "Base Image * 0.5" instead of just having the formula "0.5" to create a grey image, hence why my "Inverted" question came about, probably too much overthinking, I now see the difference int he image between TGVDenoise before and after

Let me see how I can reduce the noise now

Simon

[msmythers]

Simon

Glad you figured it out. Here is what I got after some tweaking. I'm not a fan of a smooth background unless the capture is just that good. Noise reduction is just that reduction not elimination and so many times we over do IMO. Anyway here is the screen shot.


Mike

[STAstro]

Thanks again Mike

I can now tweak TGV Denoise and get the best for this image, then I need to work out how to stop blowing out the core when doing a HDRMT, Star Masks don't seem to help at all, but I'll get to that point when I've sorted out the noise

Simon

[Tromat]

Glad to see it works for you

I'm not a fan of a smooth background unless the capture is just that good. Noise reduction is just that reduction not elimination and so many times we over do IMO.

What I like with the 50% grey mask is that you keep a slightly grainy texture, and not a fake plastic background.

[STAstro]

Just to follow up on this

Many thanks to Tromat for providing the "50% Grey" solution, I have used this now to perform an MMT against the image and this has worked better at reducing the noise than TGVDenoise, MMT is my tool of preference, and it worked well with the 50% Grey Mask

Thanks to Mike for helping me out with TGVDenoise

Now to carry on processing :-)

Simon