PixInsight Forum (historical)

PixInsight => General => Topic started by: Nicko on 2019 November 25 02:12:35

Title: Issue with registering meridian flipped debayed images
Post by: Nicko on 2019 November 25 02:12:35

Hi.

I noticed this problem when trying to register a stack of OSC pre and post meridian flipped images that had been debayed (before the registration). If I register the stack straight after calibration - all good. But after debayering not so good!

I am using a QHY128C with RGGB matrix. I have fiddled around with star alignment and debayer settings  but no success. Last resort maybe to manually flip 180 then star align.

Any ideas

THanks

John

Title: Re: Issue with registering meridian flipped debayed images
Post by: pfile on 2019 November 25 09:19:57

i don't know of any camera whose sensor is sensitive to the orientation of the camera... and although the image is formed upside-down on the sensor relative to the pre-meridian flip image, the sensor pixels themselves don't move. so the meridian flip should have no effect on whether or not you can successfully calibrate an image. even your flats will be the same pre-and post meridian flip unless you use a rotator to rotate the camera 180 degrees after the meridian flip.

also StarAlignment should have no problem with images that are rotated...

there must be some other problem going on here. can you post some of the problem frames together with your calibration masters?

rob

Title: Re: Issue with registering meridian flipped debayed images
Post by: Nicko on 2019 November 25 13:55:19

I can successfully star align frames (no calibration applied) pre & post meridian flip. Same issue, same frames (no calibration) but now debayed. Star align process seems to make no difference at all to the orientation of target frames to the ref image.

Please see link: https://www.dropbox.com/sh/y0bbz72ka39456t/AAA_yTC01bCQyRgbKheMf44Ra?dl=0

Thanks

Title: Re: Issue with registering meridian flipped debayed images
Post by: Nicko on 2019 November 25 14:22:30

Changed the Demosaicing method from VNG to SuperPixel. Star alignment after debayering appears to work. Whether this is this correct method I'm not sure but it appears to be fine.

Title: Re: Issue with registering meridian flipped debayed images
Post by: pfile on 2019 November 25 14:42:08

hmm. either should work really... i was just about to look at your frames to see if i can see anything. it could be that the stars are a little soft (focus) and using superpixel tightens them up a little... and then staralignment can latch onto them.

rob

Title: Re: Issue with registering meridian flipped debayed images
Post by: pfile on 2019 November 25 14:53:04

yeah i can see that SA is picking up on hot pixels instead of stars when debayered with VNG. to check this yourself you can load your frame and debayer it and then set the frame as a reference (using View mode), then change StarAlignment to "detected stars" mode and apply to the same frame. you'll get a new window with red +s over the detected stars. you will need to compute an STF for that image in order to see what's going on.

i think that if you go into the star detection tab of star alignment and change the following:

detection scale: 6
noise scales: 2
hot pixel removal:2
peak response: 0.88

then you'll probably have better success.

by the way the stars in frame1 are trailed enough that perhaps they will not be recognized as stars... i did all this analysis on frame 24.

rob

Title: Re: Issue with registering meridian flipped debayed images
Post by: Nicko on 2019 November 25 17:27:41

Okay, thanks Rob. Also noticed my master flat is terrible, need to fix that!