What is the best way to almost remove or shrink small stars from an image?

[Tom OD]

Hi PI,
I have an image taken with a lot of the milky way in the background. The small and tiny stars are overwheling the image. I used the Morp tool with a star mask made from the ATWT tool to reduce their brightness and size, but they are still far too many and too obvious in the image. Does anyone know of a good technique for removing the small stars completely or minimising them significantly.
Thanks Tom.

[pfile]

i suppose once you've got that mask constructed (does it really reveal all the tiny stars?) you can pull down the RGB/K curve a little bit to try to dim them down a little more...

rob

[oldwexi]

Hi Tom!
I produced a Video Tutorial to reduce the stars, the example used is not a wide angle milkyway Image, but have a look at it and maybe
there is some help for your task.
http://dl.dropbox.com/u/57910417/Howto_enhance_nebuala_without_pushing_stars.wmv

Gerald

[Tom OD]

Great guys I'll star right on this and post my results.
Thanks Tom.

[jerryyyyy]

If you go over to the Dark Side (literally) there is software called Straton De_Star that clobbers them all....

[Geoff]

Unfortunately, the stars represent reality. They are what you have imaged and the fact that they spoil the picture may be annoying, but that's what you have to live with. Removing them gives a false idea of what that part of the sky is really like.
Geoff

[jerryyyyy]

Unfortunately, the stars represent reality. They are what you have imaged and the fact that they spoil the picture may be annoying, but that's what you have to live with. Removing them gives a false idea of what that part of the sky is really like.
Geoff

Yes, I agree.  There are some techniques that remove the stars, process nebula, then put them back...

This is above my pay grade to understand.

This image used deconvolution then debloating.  If you look at the earlier versions you will see what I was trying to eliminate... my Tak blows up the stars too big as it is very sensitive: F/2.8. 

[Tom OD]

Unfortunately, the stars represent reality. They are what you have imaged and the fact that they spoil the picture may be annoying, but that's what you have to live with. Removing them gives a false idea of what that part of the sky is really like.
Geoff

Well I don't want to completely remove the stars, but I don't want them to overpower the image. I go more for the pretty picture look than the scientifically correct image, so I have no shame in dimming the background smaller ones as much as possible. In the particular image I'm processing, there is a beautiful cluster and some Alpha and Beta constellation stars that I want to leave bright to represent that part of the sky.
I have tried multiscale processing and star morph.  I have found that I can only go so far before the round stars start to go square. Maybe I need more Morph practise.
Tom.

[pfile]

are you using RBA's "contours" technique to generate the starmask for MT?

rob

[twade]

Tom,

I use the MorphologicalTransform.  It works very well in lessening the effects of a busy star field.  I typically generate a star mask and execute enough times to generate the result I'm after.  If you like, I can send you my process icon.

Wade

[cs_pixinsight]

twade, I'd like to see you process icon too.  Could you just attach it here?

Thanks, Craig

[jerryyyyy]

Hi, to stimulate conversation, these are the icons I use. 

[Tom OD]

are you using RBA's "contours" technique to generate the starmask for MT?

rob

I don't have Rogelios steps in front mf me, but I tend to use ATWT with layers 4, 5, 6, and R removed, then I clip the resultant image to get a great star mask that works on the smaller stars. I'd use a range mask o the very large stars, and the regular star mask process icon for the medium ones. Or some combination of all 3.
I don't really understand the elements section whereby you can select the 3x3, 5x5, 7x7 etc... I don't know why you would choose one over the other. I think I go for the higher values llike 9x9 as the graphic shows them rounder, and less pixelated.
I tried both methods mentioned in the post, and I thought the pulling dow on the curves gave the view I liked the most. At 200% with the Morph tool, you can see the sqaure-ish stars. At 100% I just get away with, if you're not looking for them anyway. I'll check out your process icon and see how it goes. I still need to try Geralds pixel math technique also.
Tom.

[Earl_UK]

multiple star masks and MT,s can reduce them a fair bit,

Tweak with your mask settings to get the mask you want, it will change as they shrink down I find large stars need a seperate mask, (i think i tried it on linear data for the big boys as they are only the bright and big ones there to start with)

do small MT's i use about .20

Make a new mask and repeate.

you may notice it can make holes in nebulosity, undo a few steps if you do.

[Tom OD]

Make a new mask and repeate.

[/quote]
Good tip that. I can't remember if I made a new one after each shrink. I didn't notice ringing thankfully.