Star net users?

[UlteriorModem]

Greetings fellow Star Net users!

Star net is a module for Pixinsight which does a fantastic job of removing stars from an image so that the object of interest can be enhanced leaving the stars untouched.

But there seems to be a lack of instructions or how to out there. I guess because it is so easy (just drag the process icon onto a stretched image).

I just installed it and ran it for the first time this morning. Applied to an image of the witches broom with excellent results. Stars gone, DSO remains

A few questions remain though.

What does the "Stride" variable do?

But now after I enhance the DSO how do I get the stars back?

Is there some resource out there that I missed?

[KathyNS]

The stride setting controls how it computes the starless image. It computes a single block of pixels at a time, and then repeats until it has finished the image. With a large setting, it works on fewer larger blocks of pixels. A smaller setting, it works on more blocks of fewer pixels. A large setting can produce blocky results on large stars. On the other hand, a small setting will take WAY more computer time, so it is a tradeoff. Initial documentation suggested using a stride of 64. Now, they are saying to use 128 unless you get blocky stars.

Suggested workflow is to use Starnet with "Star Mask" off to produce a starless image. You could then run it again with Star Mask on to get just the stars, but a faster way is to use PixelMath to get the stars-only image. The PixelMath expression "Original-Starless" (substituting your own image names, of course) will produce the stars-only image. Work the the starless image and the stars-only image separately until you are happy with them. Then use PixelMath with "Starless+Stars_Only" (again using your own image names) to recombine them.

Note that using PixelMath like this can affect star colours in the stars-only image. Removing a white star from a red nebula will fill in the star's pixels with red in the starless image. So when you do the PixelMath, red is subtracted from the star's white, and the star ends up cyan in the stars-only image. If you are expecting it, it is no big deal. Just resist the urge to correct the altered colour, since the inverse PixelMath operation at the end will correct it.

[Stuart]

Hey Kathy, that's great, thanks. I just downloaded it at AIC and farted with it for the first time last night.

[STEVE333]

I've had pretty good success as follows:

1) Run Starnet to produce the starless image.
2) Enhance the starless image, but, don't reduce the black level.
3) Use PixelMath to insert the stars back into the image using the expression:
max(original, starless)

Once the stars are reinserted you can reduce the black level if desired.

I haven't compared this approach to Kathy's. Hope to compare them at some point.

Steve