mosaic registration from plate solving

+1 for plate solving for really wide mosaics. I think that should speed up the registration process also for extensive mosaics.

Thanks a lot for using APP Jin Woo!

1) I'm not sure that would be the best method, I know Mabula is spending a huge amount of thought into the mosaic mathematics and I'm sure he knows about the plate solving method. I know APP is using advanced distortion correction and is the only one doing it in a good way (as far as I know from the competitors), resulting in the avoidance of mosaic problems like double stars on the edges etc. However, very wide-field is a challenge still. APP does have various projections, but the ease of use might be improved as well as the creation of mosaics for extreme wide-fields.

2) You can load those in, however APP still needs to construct a star analysis of its own to be able to work on it. Best is to reprocess the raw data in APP as you'll also benefit from its advanced calibration.

What I understand from the registration process there's the "normal" mode and the "mosaic" mode. In the latter mode, each image is compared with all others, resulting in N * (N-1) / 2 alignments when there are N images. For the normal mode this is just N-1 alignments, which makes a big difference for large N. With plate solving the first coarse mosaic alignment step can be done quite quickly I think, followed by the refinement of the positions that takes into account all the distortions that APP is so good at. It is my educated guess that mosaics with up to 300-400 frames then would become feasible, while with the current approach that really wouldn't be a smart thing to do.

Added advantages of the plate solving are e.g.

• click-of-a-button galactic, equatorial, polar, etc. projections
• identification of objects in the field of view

No rush from my side to have this feature, I'm still not ready with all exposures of my Milky Way grand mosaic 😉

That is correct (the N part), so you're proposing some kind of preprocessor method. Sounds interesting, I've asked Mabula to chime in when he has time as I'm not an expert on these algorithms.

@vincent-mod

Thanks Vincent!

I'm sure this is not just a small request and involve lot of development efforts. I think APP does excellent job with mosaic registration with star-alignment based approach; much better than the other program I'm using 🙂

My request comes from my current project; a 70 pane mosaic of the milky way taken with 105mm lens. From my experience with APP, I found (or speculate) that:

- it take very long time for the registration alone (due to the N^2 order of operations as mentioned by Ralph)

- success of mosaic registration is not guaranteed; I was successful once but after changing projection settings it fails.

- advanced projection methods seems to help but, projection error seems to propagate as number of frames increase resulting in distorted regions.

- tried to load registered images (generated by other program with plate solving method) but re-registration failed in APP; hence the request 2) without re-registration

So with APP's great local normalization correction and other features, I think that the requests would be great addition to APP especially for really big mosaic projects. 

Yes I agree, APP is doing a great job on "regular" mosaics, but could be improved still on very wide-field ones. Mabula is always improving on these area's so I'll make sure he has seen your ideas. Thanks a lot!