Back to forums

Hollow stars in star clusters

Pleiades Astrophoto PixInsight NGC6205 M13 Globular Cluster Deep Sky
17 replies•928 views
Martin S. avatar
Martin S. avatar
Hi,
about a year ago, I tried my new telescope on the M13 star cluster and after WBPP script I noticed a strange shape of some stars within the cluster. As shown below, this issue is almost invisible on non-drizzled image. However, after 2x drizzle, some, but not all, stars are completely hollow. The issue is of course even more prominent after BlurX, and the whole picture is ruined quickly.

Last few nights I acquired some more data and encountered this problem again. Luckily, this time I was able to find some old post on stargazerslounge.com, indicating that my CosmeticCorrection settings might be the problem (original post: https://stargazerslounge.com/topic/305822-hollow-star-issue/). Indeed, it was a correct hint and I was able to finally process the image without this crazy issue. All what I did was disabling CosmeticCorrections.

Hope you find it useful.
Martin

Helpful Respectful Concise Engaging Supportive
gfstallin avatar
gfstallin avatar
This is timely and interesting. I use Astro Pixel Processor to stack and noticed the same thing, though to a much lesser extent. It isn't obtrusive, but pixel peepers would notice a very slight hollowness on some of the largest stars. I'm new enough to DSO imaging (about 2 years) and processing that I had more to worry about. It definitely stands to reason that hot pixel rejection setting in any stacking software would do this. 

George
Well Written Concise
Terry Doran avatar
Terry Doran avatar
What kind of telescope?  If it’s a Newtonian, SCT or RC, could you be slightly out of focus and it’s a central obstruction?
Helpful Insightful Respectful Concise
Jim avatar
Jim avatar
I, too, ran into this on Omega Centauri. Cannot wait to try it without CC
andrea tasselli avatar
andrea tasselli avatar
Old post of mine addresses the issue: It is really just one setting that needs to be unselected when you have serious undersampling (correctly sampled or oversampled images are not affected), that of automatic detection. If you use a master dark frame instead you'll get 99.99% of the benefits without any risk, as it really should be done anyway.
Martin S. avatar
Martin S. avatar
Thanks for response.
I don't think is has anything to do with focusing. Also, different stars are hollow for different channels. This somehow implies the CosmeticCorrection issue, which was proven by reprocessing wit and without it.

Just to add some more details:
​​​​​​​
Askar FRA500
ZWO ASI533MC Pro
Optolong UV/IR cut
About 10 hours of exposure (cca 300x120s)
Rob Foster avatar
Rob Foster avatar
Todd, I think I encountered this before as well, and if memory serves, it had something to do within another area in WBPP setting having to do with interpolation mode and clamping threshold in StarAlignment; when I encountered this before, I changed up different pixel interpolation algorithms and it seemed to help.  I am not at my processing computer right now but will see if I can find those setting changes and pass them along.  

Sounds like your CC fix did the job, but would be interesting to see if you restored CC but lowered the Clamping threshold in StarAlignment or choose another Pixel interpolation algorithm whether it would go away as well. 
CS,
Rob
Supportive
Tony Gondola avatar
Tony Gondola avatar
This can also happen if the stars are clipped. If that's the case then nit's in the original data, just hard to see. Processing simply makes it more visible. You might try a line contour plot on the original data and see what it looks like. If you get flat tops, the original data is clipped.
Helpful Concise
Jeff Kisslinger avatar
Jeff Kisslinger avatar
How did you stretch the image? I notice these same artifacts when I use masked stretch and Arcsihn stretch. I just posted images of M13 and M92 and had no issues with non-drizzled data stretching with just Histogram Transformation. In these particular instances I did use darks with the auto correction feature enabled in WBPP.
Martin S. avatar
Martin S. avatar
All the screenshots are just temporarily stretched using linked STF AutoStretch.
I went through some other posts showing the clipping, but it looked a bit different.
And yes, I used somewhat old darks. I definitely should redo my darks library.
Rob Foster avatar
Rob Foster avatar
Following up on Tony's point, HSV Repair Script can help if they are indeed clipped. I've gone that route as well! That was more of an issue with DSLR imaging than a cooled camera, but take a look at the stars in the linear state and see what the readout mode is showing for K value - anything over .8 is clipped and might benefit.   However, the universal finding of so may cored out samples makes me think it is more likely another setting in WBPP - 
CS,
Rob
Robin Livermore avatar
Robin Livermore avatar
Yes it's CosmeticCorrection. The tiniest stars often have a single bright pixel in the centre - if you use auto detect CC can detect these as outliers and remove them!
Helpful Concise
Niels_L avatar
Niels_L avatar
Many suggestions here - I had the same issue. Read @andrea tasselli  answer, it is the correct way to do CC (on your masterdark)  Never had the issue again.
quantOnaut avatar
quantOnaut avatar
Martin S.:
Hi,
about a year ago, I tried my new telescope on the M13 star cluster and after WBPP script I noticed a strange shape of some stars within the cluster. As shown below, this issue is almost invisible on non-drizzled image. However, after 2x drizzle, some, but not all, stars are completely hollow. The issue is of course even more prominent after BlurX, and the whole picture is ruined quickly.

Last few nights I acquired some more data and encountered this problem again. Luckily, this time I was able to find some old post on stargazerslounge.com, indicating that my CosmeticCorrection settings might be the problem (original post: https://stargazerslounge.com/topic/305822-hollow-star-issue/). Indeed, it was a correct hint and I was able to finally process the image without this crazy issue. All what I did was disabling CosmeticCorrections.

Hope you find it useful.
Martin


During my 6 year long astrophotography journey experimented with all common types of telescopes but never managed hollow stars from stacking, who introduced this weird algorithm 😆 ? From my perspective it looks like out of focus reflector showing the secondary mirror maybe with stacking process it amplified and became visible and further processing is just bringing out the real disaster behind the raw images? With 11” reflectors and Front Glass corrector plate the stable focus can be challenging over night!

 

Why drizzle shows “hollow stars” more clearly

 

  1. Drizzle (especially 2x) reconstructs the image at higher resolution:
    • It doesn’t invent detail but redistributes flux more precisely using sub-pixel shifts between frames.
    • If the stars are slightly out of focus, the star’s light is already slightly donut-shaped or flat-topped in the raw data — drizzle will magnify that profile, making the core appear “hollow”

  2. PSF behavior in slightly defocused systems:
    • When a star is out of focus, the point spread fNOunction (PSF) becomes a bowl or donut rather than a tight Gaussian.
    • Drizzle samples that more finely, so instead of a soft blur, you get a resolved dip in the center

  3. Large central obstruction of SCTs (like C11):
    • The Airy disk of SCTs already has a pronounced ring-like structure.
    • Combine that with minor defocus or seeing, and drizzle can visually amplify the central dip

  4. Undersampling before drizzle:
    • If the camera had undersampled stars before drizzle (e.g., large pixels or short focal length for the seeing), the original stellar profile was not properly recorded.
    • Drizzle tries to reconstruct sub-pixel information, but with poor input data (like flat-topped or too-small stars), it can produce artifacts

I have overprocessed your NOdrizzle Version and you can clearly see tiny black dots on the stars Center which have been amplified by drizzle process later.

Helpful Insightful
Kevin Morefield avatar
Kevin Morefield avatar
andrea tasselli:
Old post of mine addresses the issue: It is really just one setting that needs to be unselected when you have serious undersampling (correctly sampled or oversampled images are not affected), that of automatic detection. If you use a master dark frame instead you'll get 99.99% of the benefits without any risk, as it really should be done anyway.

And there are other issues with Autocorrection.  If you see a little line between two close but resolved stars, it may be created by the dark autocorrection in CC.  I use master darks always for CC.  3 sigma light and dark.  But I also use Autocorrect at 6 sigma just in case.  Probably useless but it seems to avoid artifacts at that level.

Kevin
Martin S. avatar
Martin S. avatar
...and here is the final image. Well, at least for tonight ;-) Thanks for all the comments here!
https://www.astrobin.com/0lv4jn/
Tony Gondola avatar
Tony Gondola avatar
That's a good looking M13, good resolution and color.
Well Written
Ed Magowan avatar
Ed Magowan avatar
This effect due to 'Cosmetic Correction' has been a bugger since the early days of DeepSkyStacker.