I'm about to shoot an all-sky mosaic

Hi @ddnum,

Indeed, APP can register widefield data 😉 with a very large field of view. For example this 180 degree mosaic, shot at 14mm focal length:

In your case, the field of view will be larger than 120 degrees, since you mention an all-sky mosaic.

In that case, you must use the calibrated projective registration mode in 4). The calibration enables a different data projection than the default rectilinear projection. A rectilinear projection will never work with a field of view larger than 120 degrees.

I was wondering what are the optimal alt-azimuth values to specify to get the best results from APP.

There aren't, but I would recommend enough overlap between the mosaic panels, with more overlap, the result will be more robust. So 20% overlap will give you a much higher chance of a good result than 10% overlap 😉

Another issue is that I don't really know how to uniformly spread alt-azimuth values uniformly over a sphere. But I'll figure that out. In the meantime I might just rotate azimuth from 0 to 360 and altitude from 0 to 90. It will get quite squished near the zenith. Is that a challenge for APP?

APP can handle that, this is a 30 panel mosaic surrounding polaris (data courtesy of Scott Rosen @scott_rosen ) shot with 2 different camera's:

Do let me know if you need assistance when you are starting to process the data 😉 . I have been working on a big upgrade of the registration engine as well to also further improve and speed-up mosaic registration and to add more data projections like sperical.

Kind regards,

Mabula

Thanks for the reply. I'm going to shoot 21 images at the vertices of an icosahedron. I'm doing the calculations to see if there will be enough overlap.

Where the images overlap does APP integrate the pixels? If each panel is a stack of 4 images does the overlapped area have the SNR ratio of 8 images? If that is the case then I can really go hard on the overlapping area knowing that it increases my SNR ratio so no pixels are wasted.