Reference frame is ...
 
Share:
Notifications
Clear all

Reference frame is distorted during registration  

  RSS

(@rainer-raupach)
Molecular Cloud Customer
Joined: 1 year ago
Posts: 3
October 22, 2021 20:53  

Hi,
I have a fully registered set of RGBHO stacks with telescope A. Now, I want to add SII data taken with telescope B.

My expectation was, that I can simply register the new SII integrated stack to one of the old channels (set as the reference, let say choosing green) where the reference will potentially be extended at the edges to obtain the same final size, but, left undistorted otherwise by definition.

Registration works perfectly using dynamic distortion (same optics unchecked), however, BOTH images are distorted during registration process. The only solution with this behavior would be to register all channels simultaneously again with the problem that I cannot use already calculated starless version any more.

Is there a way to force the registration into leaving the reference frame exactly as it is (up to padding with zero at the edges to enlarge the common matrix) in combination with dynamic distortion?

Thanks for any idea!
Rainer

This topic was modified 1 month ago 2 times by Rainer Raupach

ReplyQuote
(@vincent-mod)
Universe Admin
Joined: 4 years ago
Posts: 3961
October 22, 2021 21:25  

Not sure I fully understand your goal, but what happens if you don't use dynamic distortion correction?


ReplyQuote
(@rainer-raupach)
Molecular Cloud Customer
Joined: 1 year ago
Posts: 3
October 22, 2021 21:46  

@vincent-mod
Thanks for the quick answer, Vincent!
I've also tried that already. Without dynamic distortion correction registration quality is very bad because the two optics are too different. Dynamic distortion correction is key to registration sucess but it unfortunately modifies the reference frame as well (beyond simple shifting or padding which would be acceptable).


ReplyQuote
(@mabula-admin)
Universe Admin
Joined: 5 years ago
Posts: 2747
October 22, 2021 21:58  

Hi Rainer@rainer-raupach,

I would not be surprised that the problem is this :

that I cannot use already calculated starless version

So you are registering the SII data of telescope B to star reduced/starless stacks of telescope A. Maybe too little stars are left to actually make a robust optical distortion calculation in your case. Or the stars that are left are too sharp and unnatural for robust calculations?

I will be happy to have a look at the data to see which would be the case. Do you want me to check your data?

"Is there a way to force the registration into leaving the reference frame exactly as it is (up to padding with zero at the edges to enlarge the common matrix) in combination with dynamic distortion"

Well, the main strong point of APP in all of this is that optical distortion correction is applied to the reference frame as well, where other applications only simply deform the rest of the frames and leave the reference intact... It is not difficult to make your request possible, but it will lead to an extra setting in the registration engine which from a physics and maths point of view should not be used I think, so I need to think about that.

The Field of View of the Reference can be maintained as it is, by setting the composition mode in 6) Integrate. You will need to set it to Reference instead of Full 😉

Mabula


ReplyQuote
(@rainer-raupach)
Molecular Cloud Customer
Joined: 1 year ago
Posts: 3
October 23, 2021 09:40  

Hi Mabula @mabula-admin,

thanks a lot for the quick answer. I'm not trying to register starless images of course (I'm a physicist working in the field of signal theory and image processing😉). I just want to reuse starless versions based on already registered stacks with stars.

I agree that in general dynamic distortion correction gives the best results if applied to all frames. APP does a really great job here even if the optics are completely different. I could not register the two optics satisfyingly with any other tool I tried.

However, for this special task of adding channels later to already registered sets without the need of redoing post-processing on the old data it would be optimal to get the behavior (A,B) -> (A,B') with A the reference also in combination with dynamic distortion correction. i.e. leaving the reference intact.

Yes, unfortunately, this would be an option rather than the default behavior and require an additional switch possibly causing new complaints because it's an additional source for operating errors. But I would really be happy having this possibility 'keep reference intact'.🙂 

Thanks again, Rainer

This post was modified 1 month ago by Rainer Raupach

ReplyQuote
Share: