Help with pasting stars on comet image

[Jason Tackett]

Hello everyone,

I am working on an image of Comet C/2012 K1 PANSTARRS that I shot last month with the goal to have the stars untrailed. I’m having some trouble combining the stars onto the comet image using PixelMath and a star mask. The pasted stars become very bright as shown in the bottom-most image below.  Is it possible that I am using the star mask incorrectly or maybe I missed a fundamental step?

I have been following Harry’s excellent comet stacking tutorial where he shows how to make two images – one stack registered to the stars and the second stack registered to the comet. After using linear fit to make the backgrounds match, PixelMath is used to paste the star stack image onto the comet stack image through a star mask. When I do this, the brightness of the pasted stars increases dramatically and the stars no longer look like the stars in the star stack image.

Here are my steps. Note that I used sigma clipping because there are only 9 subs and I used CloneTool to remove the large halo around the brightest star and the galaxy in the comet_stack image.

1. Calibrate subs w/darks, bias, flats
2. Register subs to stars (StarRegistration)
3. Register subs to comets (CometRegistration)
4. Integrate star-registered subs to make star_stack (sigma clipping [low, high: 4.0, 2.0])
5. Integrate comet-registered subs to make comet_stack (sigma clipping [low, high: 4.0, 1.5])
6. Remove remainder of large star halo and galaxy from comet stack (CloneTool)
7. Match backgrounds (LinearFit, reference: star_stack)
8. Create star mask from star_stack (StarMask)
9. Stretch star mask to brighten fainter stars (HistogramTransformation)
10. Remove comet core from star mask (CloneTool)
11. Apply star mask to comet_stack
12. Paste stars onto comet_stack (PixelMath: expression: star_stack)

The attached images show the star_stack, comet_stack, star_mask, and comet_stack after pasting stars. The close-up image shows how much brighter the stars look than in the star_stack image.

FWIW, this is a short sequence of 9 x 90s exposures with a Canon 1000D @ ISO1600 through a 127 mm apochromatic refractor.  Fit files for for these images can be found here: https://www.dropbox.com/sh/57bahuk99zw7qil/AABMOOTBpZLpjas7Z3SEBn74a

Many thanks for any advice on how to tame these stars…and big thanks to Harry for the great tutorials. They are incredibly helpful.

Best,
Jason

[FunTomas]

Hi Jason.

Your star_stack image and comet_stack image probably doesn't have the same brightness. Tabs on your images have the green line, these images have only screen transfer applied. They are not stretched. Try to apply LinearFit  to one of these images with second image as reference. If nothing will help to you, please share somewhere your fits (star_stack and comet_stack).

Tomáš

[Jason Tackett]

Hi Tomáš,

Thank you very much for your reply. Previously I applied LinearFit to the comet_stack image using star_stack as a reference and then confirmed the median value of a small preview of the background in both images was nearly identical prior to pasting the stars with PixelMath. If anyone would like to examine the .fits files for star_stack and comet_stack, they can be found here:

https://www.dropbox.com/sh/57bahuk99zw7qil/AABMOOTBpZLpjas7Z3SEBn74a

Thanks again,
Jason

[MikeOates]

Jason,

To reduce those artefacts around the stars, try to modify the star_mask, either using convolution to soften the edges or use MorphologicalTransform to enlarge the stars in the mask, you may also need some convolution as well.

Mike

[Jason Tackett]

Thanks for the suggestion, Mike. I will give it a shot and report back.

Jason

[FunTomas]

Hi Jason.

Sorry for my later answer. Your images have different brightness. This is important for good background around the stars. You need to match them. Secondary.. What kind of image integration are you used? Use more restrictive settings for high values. Today I try to create image and send it to you.

Tomáš

[Jason Tackett]

Hi Tomáš,

Many thanks for the examples you sent and for helping me with this project. Here is the final processed image:

http://www.astrobin.com/105937/

You are right that my mistake was that the background values did not match between the two images. To resolve this I used the Statistics process and Histogram Transformation to independently set identical median background values for the comet_stack and star_stack images. My complete processing steps are listed in the image description on my Astrobin page for anyone interested.

Also, thanks for the suggestion, Mike - I spent quite a bit of time making my range mask to select the stars and galaxy. Expanding and softening the edges as you recommended helped quite a bit.

The image came out decent, mostly since I only acquired 13.5 minutes of exposure time that night before the moon and clouds decided to shut down the party. Still, it is a lovely combination of objects between the comet, the galaxy NGC 3877 and the bright orange star Chi Usae Majoris (which I could have controlled the bloating on a little better).

Thanks again for the help. I really appreciate you taking the time to set me on the right track.

Best,
Jason

[MikeOates]

Jason,

That's great, glad I could be of help. You did well with such short exposures.

Mike