Controlling Star Profiles with Star Shrink

[Ken Crawford (KenC)]

We all have many different ways to control the color, size, and profiles of stars in our astro-images. For several years I have been using a prototype of a Photoshop plugin filter written by Russ Croman (Author of GradientX) called Star Shrink. I thought I would share a tutorial I just completed on the basics of using this amazing but easy to use filter. I believe you may find this interesting . . . .

http://www.rc-astro.com/resources/St.../tutorial.html

Kindest Regards,

Ken Crawford

[multiweb (Marc)]

I was sold after watching your vimeo tut Ken but on first real world use it seems to target a lot of nebulosity unrelated to stars. How does it detect star? See distortion in flame and hh. Masks are a necessity but then it defeats the purpose a little for rich star fields.

[rcroman]

Originally Posted by multiweb

I was sold after watching your vimeo tut Ken but on first real world use it seems to target a lot of nebulosity unrelated to stars. How does it detect star? See distortion in flame and hh. Masks are a necessity but then it defeats the purpose a little for rich star fields.

Hi Marc,

StarShrink tries to detect stars by looking for sharp changes in brightness. It generally does a pretty good job, since stars are generally where the sharpest changes in brightness in an image are. But it can pick up on non-stellar objects as well, especially areas of high contrast, and especially at higher radius settings. Alas, there's no such thing as a perfect filter. The aim with this one was to enable a mostly-automatic method of tightening star profiles in a way that was previously quite labor-intensive and difficult to get consistent results.

Ken's tutorial shows some easy ways to control which parts of the image are affected.

Best,

Russ

[multiweb (Marc)]

Originally Posted by rcroman

Hi Marc,

StarShrink tries to detect stars by looking for sharp changes in brightness. It generally does a pretty good job, since stars are generally where the sharpest changes in brightness in an image are. But it can pick up on non-stellar objects as well, especially areas of high contrast, and especially at higher radius settings. Alas, there's no such thing as a perfect filter. The aim with this one was to enable a mostly-automatic method of tightening star profiles in a way that was previously quite labor-intensive and difficult to get consistent results.

Ken's tutorial shows some easy ways to control which parts of the image are affected.

Best,

Russ

Hi Russel, thanks for the clarification. So the trigger for selection is any sharp boundary between bright and dark? Probably why it selected the dark ridge inside the flame. Will play with the settings again and be mindful to use a mask as not to affect any other part of the photo I'm working on.

[rcroman]

Originally Posted by multiweb

So the trigger for selection is any sharp boundary between bright and dark?

That's correct... any sharp boundary will tend to be picked up a bit by the filter. It's more pronounced at higher radius settings.

One day I hope to come up with the perfect star-selection algorithm, but it's a tricky problem. Even the professional astronomer's image processing software mistakes non-stellar objects for stars sometimes.

[multiweb (Marc)]

Originally Posted by rcroman

That's correct... any sharp boundary will tend to be picked up a bit by the filter. It's more pronounced at higher radius settings.

Anything within the radius setting gets 'pinched' towards the centre of the selection like a warp. Would there be a way to identify a star as a sharp change in luminosity but in a closed looped edge rather than a linear boundary? What I'm getting at is possibly generating a temporary mask across the whole picture as to not affect all edges, stars only.

[rcroman]

Originally Posted by multiweb

Anything within the radius setting gets 'pinched' towards the centre of the selection like a warp. Would there be a way to identify a star as a sharp change in luminosity but in a closed looped edge rather than a linear boundary? What I'm getting at is possibly generating a temporary mask across the whole picture as to not affect all edges, stars only.

Perhaps. I've played around with various approaches... none so far have been effective enough to be better than the usual masking techniques.

[multiweb (Marc)]

Originally Posted by rcroman

Perhaps. I've played around with various approaches... none so far have been effective enough to be better than the usual masking techniques.

Fair enough.

[Ken Crawford (KenC)]

Originally Posted by multiweb

Fair enough.

For complex images a good star mask really makes it work well . .

I was able to use Shrink on this 3 panel mosaic with lots of nebula around with an a star mask. I use the same star mask to control stretching around the stars, color, and in this case applied OIII - SII - Ha to an LRGB without adding that data to the stars.

http://www.imagingdeepsky.com/Nebula...00/NGC7000.htm

Kindest Regards,

[NightCal]

I've just started using the trial version and so far I'm impressed.
I was wondering at what stage in the image processing cycle would it be best to apply this plug-in? Given that the star recognition algorthimn is looking for sharp transitions, would I correct in assuming that (in order to reduce the risk of it selecting non-stellar object) it would be best to apply the plug-in after any noise reduction, but before any sharpening techniques?

[bilgebay (Sedat)]

A great tool for sure! Thank you Russ... and thank you Ken for the tutorial. How does Starshrink compare to Pixinsight's Morphological Transformation tool ?

Clear skies

Sedat