Pixinsight Image Registration and camera orientation

[billhinge]

I have been taking some images of NGC6543 via a remote telescope and just over a month ago there was a fault with the camera (SBIG STL-11000M @ 4008 x 2672 pixels )

I had a set of images taken before the camera fault and I took some more last night, however the camera orientation seems to have changed by 90 degrees following repair

When I register all my images in PI I get what appears two different image orientations overlaid

Question is, is PI intelligent enough to reorient the images before doing alignment or does it reject half the images that are 90 degrees out of orientation - should I manually rotate all the images to the same orientation?

see image below as an example showing a mixed set of registered images after stacking

(I assume I'm going to end up with a stacked image that is closer to square than rectangular)

[The Sailor]

My thoughts would be to keep both orientatined photos as seperate images, sort out what you want to do in PI and then rotate whichever one you wish and then redo them in PI. Maybe someone will have a better reply but that is how I would do it, a bit of cropping will have to be done so square them off .

Jim

[GlassWalker]

PixInsight does rotate as necessary during registration. It can even scale the image if you use different sources.

If you have two sets of data 90 degrees rotated, then the image above looks within expectations. Your sensor isn't square, so when the image is overlapped, you can get different looking bits if the sky conditions weren't the same.

[billhinge]

Thanks guys, sounds like good answers. I claimed and got a refund for the inconvenience so I'm OK, and I'm not missing anything of importance by getting a square final image

[rfdesigner]

Can you rotate one batch.. then align and stack?

I believe IRIS could get this right (which has been my program of choice to date), so I would expect PI to be able to do this

Derek

[James]

Question is, is PI intelligent enough to reorient the images before doing alignment or does it reject half the images that are 90 degrees out of orientation - should I manually rotate all the images to the same orientation?

PI will cope with that just fine. I've thrown in subs that are at 90 degrees out, 180 degrees out and at a completely different scale and PI just chews it all up and spits out the finished product without even blinking..

James

[billhinge]

Thanks James

So are you saying I don't have to use the same telescope/camera combination?, that a similar combination would work

would I need to pre-process (rescale) them first?

ie say

combo 1 is

Array: 4008 by 2672 (10.7 Mega pixels)

FOV: 28.2 x 42.3 arc-mins

combo 2 is

Array: 3072 x 2048 (6.3 Megapixels)

FOV: 37.41 x 24.94 arc-mins

[GlassWalker]

I've mixed 12MP and 18MP images without major problem before. Same sensor size, same optics, different wavelengths, different sensor MP count. I also aim by eye so I'm never quite pointing at exactly the same bit of sky anyway... and it copes with that too. Recent example below where you can see the three frames overlapping the region of interest and the rotation between them. The OIII was with the 18MP, the Ha and SII were at 12MP.

I did make a mistake before and tried registering wildly different scale FoV (over 2x difference), and in that case it did get confused, so there are limits. In extreme cases like that, you might need to crop and/or scale to get them in the same ball park first. But generally keep it sane and it should be ok.

[billhinge]

Thanks for the info, I didn't realise it was that robust, I assumed everything had to be pretty close, I'm sure I tried some pretty ropey short duration shots with a camera on an unguided tripod once and got RANSAC failures.

It was quite a while back so I may have misinterpreted something