Artifacts after Star Alignment ?

[Ioannis Ioannou]

Hello forum,

For years, I follow the same procedure for pre-proccessing: defect map, calibration with overscan+bias(or superbias recently)+flat, registration and finally integration. This worked fine for the last 4-5 years, both with my previous camera (QHY IMG2P) and the current one (QHY22). Almost never had a problem with this procedure. Now, after months of inactivity, I'm trying to process images acquired last summer, and I have some very strange results after registration (Star Alignment). FYI my parameters for Star Alignment are the default, I only define a reference frame and I select that I want to create drizzle files, nothing else.

Take a look on the snapshot, what the heck are these artifacts around the stars? They are not there up to the calibration stage, they appeared only after Star Alignment. They appeared no matter if I'm using Pentagons or Triangles, on all registered frames except of the reference one, and of course they are apparent after integration (no matter what rejection algorithm I'm using). I have 12 frames per RGB, acquired with maxim, dithered, bin2x2 and the problem exists on all three channels.

Strange enough, if I register+integrate (align+stack) with Nebulosity 3, no similar artifact appears. Did not tried with another program, but I think it should be something in Star Alignment parameters. Something changed in the Star Alignment and I missed it ?

(PI 01.08.02.1098 on Win7/64bit, as far as I can tell fully patched)


Thanks in Advance

[georg.viehoever]

It is probably related to the Pixel Interpolation algorithm used. Try to select the bilinear algorithm, or play with the clamping threshold.
Georg

[Ioannis Ioannou]

Phew! Georg that was FAST ! And useful as always 
Bilinear is ok, even though the result is not as sharp as the original - but there are no artifacts, so I guess I have to play with interpolation and see which parameter fits better to me.

(BTW when this changed ? I never had to bother with the interpolation algorithm before ! )

Thank you

[georg.viehoever]

...
(BTW when this changed ? I never had to bother with the interpolation algorithm before ! )
...

Probably in 2012, see http://pixinsight.com/doc/tools/StarAlignment/StarAlignment.html#description_006 . Lancos interpolation seems to cause his kind of ringing sometimes.
Georg

[Ioannis Ioannou]

I see. I tried clamping also with good results, I'm playing with the parameters now to find the optimum in my case.
I'm guessing I never noticed it before because I rarely do binx2, my binx1 L is ok with the default.

Again thank you Georg