Page 1 of 1

Wipe and very faint backgroud gradients

Posted: Wed Jul 15, 2015 9:53 am
by akeru
Rogelio Bernal, a great astrophotographer know for a lot of us, shared a removal gradient tutorial in his web.
http://www.deepskycolors.com/archivo/20 ... ng-ve.html

I would like to know if it is possible with startools make a wipe processing similar to this DBE in pixinsight because in some circumstances we are interested in preserving details in some regions of the picture. Thanks a lot.

Re: Wipe and very faint backgroud gradients

Posted: Thu Jul 16, 2015 12:21 am
by midwayexpress
Interesting. Personally I have been finding that startools actually does a much better on background removal than pixinsight, especially in 1.4.

Re: Wipe and very faint backgroud gradients

Posted: Thu Jul 16, 2015 4:19 am
by admin
I would like to know if it is possible with startools make a wipe processing similar to this DBE in pixinsight because in some circumstances we are interested in preserving details in some regions of the picture. Thanks a lot.
Definitely! StarTools should already do a good job by itself in most cases, however, everything described in the tutorial has a StarTools equivalent;

Rogelio's step 1. Use AutoDev to see what's in the image. You can also set 'Temporary Autodev' on in the Wipe module.

Rogelio's step 2. Instead of 'placing samples', you do the opposite in StarTools; create gaps in the mask where you don't want Wipe to sample the image. So if you know you have nebulosity in an area and you believe Wipe is removing it, simply mask it out (for example with the Lassoo tool). With Wipe this is something that is fairly rare, but it may happen if you choose high 'aggressiveness' values. This procedure is performed in this video to preserve M8's nebulosity a bit better.

Rogelio's step 3. Is unnecessary (it's taken care of automatically).

Hope this helps!

Re: Wipe and very faint backgroud gradients

Posted: Thu Jul 16, 2015 6:55 am
by akeru
Thanks a lot Ivo, it was I supposed. Really wipe does a very good work removing gradients. In this procedure Rogelio apply a different weight to the gradient mask, do you plan add this feature in future versions of startools? About your suggestion, applying a invert mask before a wipe we can select or unselect a region but it is not possible select a region with less aggresivity. If I would like do this could i do the wipe process two times with different mask?

Regards,
Roberto.

Re: Wipe and very faint backgroud gradients

Posted: Thu Jul 16, 2015 7:47 am
by admin
akeru wrote: In this procedure Rogelio apply a different weight to the gradient mask, do you plan add this feature in future versions of startools?
I'm not entirely sure what you mean by "applying a different weight to the gradient mask". Can you elaborate?
About your suggestion, applying a invert mask before a wipe we can select or unselect a region but it is not possible select a region with less aggresivity. If I would like do this could i do the wipe process two times with different mask?

Regards,
Roberto.
The reason why Wipe tends to do a better job than DBE in most instances - and especially in the case of Rogelio's data - is because the 'aggressiveness' defines a size boundary, below which "gradients" are left alone (which would actually solve Rogelio's problem from the get-go!). In Wipe's algorithm, gradient removal is not just dictated by the samples in the mask - they are a secondary consideration. Sample spacing has no real effect either like it does in DBE; de-selecting a region will merely exclude that region when looking for gradients to construct the global gradient model, however gradient subtraction is still performed on that region with the same aggressiveness consideration based on any detected gradients outside those areas (we're just hiding further gradients from Wipe
s detection algorithms in those excluded areas). The latter means that if Wipe detected a gradient that is bigger than the area you excluded, it will still model and subtract the bigger gradient from that area.

Again, aggressiveness is a parameter that specifies the likely size of the gradient (below which undulations and other, 'faster' gradients are considered real detail). Vignetting is the only typical source of a gradient that can cause fast undulation towards the corners. That's why a 'corner aggressiveness' can be specified. If you need local aggressiveness controls that means you picked the wrong aggressiveness setting for the rest of the image... :)

Re: Wipe and very faint backgroud gradients

Posted: Thu Jul 16, 2015 9:59 am
by akeru
Ok, I think it was clear now for me, thanks Ivo.

P.D: I want use "weight" as "force" or "aggresivity" in my sentence, excuse me, I have problems with the translation :doh: