Closed timegate closed 1 year ago
I see. This problem is most likely due to all the extra space in the target image that has no correspondence with either of the sources --i.e. the field of view of both is quite different.
I think masking the region of interest (masking out the extra space) could be a solution. This is related to #90. I'll try implementing this ASAP and we can try again with masking.
Wow, thanks for fast and kind reply. I got the point exactly. Thanks so much.
Umm.. I hope that we find a way to align 2 images regardless of fov. I found that astrometry finds WCS well regardless of fov, so maybe it would be nice to leverage the two WCSs found so that the aligning works regardless of fov.
Maybe 1. Astrometry -> 2. Astroalign with known stars (https://astroalign.quatrope.org/en/latest/tutorial.html#if-you-know-the-star-to-star-correspondence) could be one way. How do you think about this way?
Sure, if you want to involve WCS, I would recommend solving it with astrometry.net and then use something like reproject to do the mapping.
Extra note: Once you have the WCS, there are better tools to do the re-mapping, reproject being one. Astroalign may not be the best in that case.
Thanks for enlightening me. I learned a lot and I'll give it a try. Thanks so much!
I tried to align below 2 images, but I get an error "astroalign.MaxIterError: List of matching triangles exhausted before an acceptable transformation was found." source target
I thought it was because it has too much candidates, so I tried a source image with fewer stars source2
But the same error happened again (changing source <-> target also showed the same error).
Could you please explain why it doesn't work? or is there any workaround?