Alignment produced double image

[heyjp]

I've processed a "goodly" number of images in pixinsight, but had something new happen today.

I shot a series of frames of NGC 4565, the Needle, and when processed thru WBPP, the integrated image was a double-image. Blinking the registered images, they had 2 different primary alignments

I took the images and ran them manually with StarAlignment, and same thing. The output "registered" images had a couple of different alignment points.

Any idea why this happened? (stacked image attached)

Jim

 

[cloudbait]

I've processed a "goodly" number of images in pixinsight, but had something new happen today.

I shot a series of frames of NGC 4565, the Needle, and when processed thru WBPP, the integrated image was a double-image. Blinking the registered images, they had 2 different primary alignments

I took the images and ran them manually with StarAlignment, and same thing. The output "registered" images had a couple of different alignment points.

Any idea why this happened? (stacked image attached)

Jim

View attachment 22943

Were you using CosmeticCorrection? Hot pixels can produce this kind of (mis)alignment.

 

[Mike1485]

Have you checked the image used as the reference for registration - did something go amiss in the capture of that image so it shows this doubling?

 

[heyjp]

Were you using CosmeticCorrection? Hot pixels can produce this kind of (mis)alignment.

Hi Chris,

Yes, I was using CosmeticCorrection, so I went back and disabled it. No change. I also bumped up some of the parameters related to rejecting hot pixels to make it reject 2x2 pixel "stars" rather than 1. Still no difference.

I also "blinked" each of the images and definitely noted that at some point, maybe a cloud obscured my tracking stars and it resettled with everything shifted slightly. But still, I would expect registration to account for that (and it has many times in the past).

Jim in Boulder

 

[heyjp]

Have you checked the image used as the reference for registration - did something go amiss in the capture of that image so it shows this doubling?

Hi Mike,
Good thought, no I have not checked the registration image. Although 1) I ran again with a manually selected reference image while turning off "auto" for reference selection. No change. And 2) I have blinked all the raw images and all look reasonable, no double images. Although I think a cloud obscured my tracking stars for awhile, and when it resumed, everything was shifted a few pixels, but I would expect registration to take care of that!

 

[cloudbait]

Hi Chris,

Yes, I was using CosmeticCorrection, so I went back and disabled it. No change. I also bumped up some of the parameters related to rejecting hot pixels to make it reject 2x2 pixel "stars" rather than 1. Still no difference.

I also "blinked" each of the images and definitely noted that at some point, maybe a cloud obscured my tracking stars and it resettled with everything shifted slightly. But still, I would expect registration to account for that (and it has many times in the past).

Jim in Boulder

I'd blink through the registered frames. That's probably more diagnostic of whatever is happening in the registration process.

 

[heyjp]

Chris, I blinked the registered frames and immediately saw the star patterns were mobile and shifting some. Then noticed that there were a smattering of individual pixels that were dead-steady. Looks like these hot pixels were winning the registration battle. I re-ran WBPP bumping up the pixel rejection boxes to "2". But no change. I've run it with and without Cosmetic Correction.

I thought CC was specifically to get rid of hot and cold pixels, yes?

What settings do you typically use in the CC setup box?

JIM

 

[fredvanner]

What settings do you typically use in the CC setup box?

I usually select auto detect hot pixels sigma=3.
(I've never had any problems with cold pixels)

 

[cloudbait]

Chris, I blinked the registered frames and immediately saw the star patterns were mobile and shifting some. Then noticed that there were a smattering of individual pixels that were dead-steady. Looks like these hot pixels were winning the registration battle. I re-ran WBPP bumping up the pixel rejection boxes to "2". But no change. I've run it with and without Cosmetic Correction.

I thought CC was specifically to get rid of hot and cold pixels, yes?

What settings do you typically use in the CC setup box?

JIM

Which pixel rejection boxes? Pixel rejection normally occurs at the integration stage, after registration. There is a hot pixel adjustment in the WBPP registration stage, but I've never used it. CC works fine for me. Same setting as Fred suggests- auto detect hot sigma 3. If that still leaves too many, make the sigma value a little smaller (not larger, as intuition might suggest).