Green scan-line artifact across some bright and saturated stars

[Mark Lowenthal]

ASTAP Version 2025-06-05, Intel Mac version
I just started testing the new version and have noticed a problem on some saturated stars. Its a single green horizontal line that looks like a single pixel line or an anti-aliased single pixel line crossing over the bright core of the star (not crossing the star center though), or as an single pixel artifact on both sides of the bright star, but not across the bright star. Its happened to at least one saturated star in every OSC image I have processed with the ASTAP 06-05 version so far. The green pixel values are typically much higher than 100,000 or even 200,000, while the rest of the saturated bright star color values max out at well less than 100,000.

It looks to me as if it's some sort of bad interaction between the AstroC debayer algorithm and the new reverse mapping algorithm implemented in the new ASTAP version. Color

This attached fits and tifs show cropped examples of several stars in 3 different images, which were stacks of 46, 60 and 222 images. The tifs are stretched so the green lines are apparent and the fits are the is the raw data from the ASTAP stack cropped using Fitsworks in order to retain the original fits header data that ASTAP wrote. Stack method settings is a snapshot of the stacking settings I used.

If you need more info, let me know. I like the new algorthm! It reduced rms noise in my images by more than 10%. The only problem I see issue with saturated stars.

[han.k]

Hi Mark,

I will look into it.

Cheers, Han

[han.k]

It is related to threading. In the latest version the image is split in sections to speed up. The problem occurs at the boundary between two threads....

[han.k]

I think I found the problem. Try the new version 2025.06.15. Tell me if it is gone.

Cheers, Han

[Mark Lowenthal]

Thanks for the quick work on this!

I think 9 out of 10 images I reprocessed to get lower noise had at least one bad saturated star that exhibited the problem in v2025-06-05. So these corruptions are statistically likely to occur.

I just processed two images through v2025-06-05, both of which showed examples of a corrupted saturated star. Then, I reprocessed both images using the same sets of subs with v2025-06-15 and saw the saturated stars that had been corrupted looked good. I also check through the other saturated stars in the these two images to make sure I hadn't missed a bad star and I then subtracted the v2025-06-15 image versions from v2025-06-15 image versions to see if any new corruption showed up that wasn't in the images produced with v2025-06-05. I found no new examples of corruption in the v2025-06-15 output.

I think this is probably enough prove that issue has been resolved!