Can you give a ball park idea of your definition of large scale?
From the point of view of a process such as StarAlignment, large scale means that large-scale geometric distortions become difficult to deal with. This may happen with just two mosaic frames or with sixteen frames, depending on image scales and optical distortions. Very large mosaics pose the problem of projecting a sphere on the plane. When the deformations caused by all of these factors become large, a mosaic construction tool based on astrometric solutions and a predefined projection system, such as
MosaicByCoordinates, can be much more efficient and accurate. Astrometry-based image registration has the additional advantage that it does not need overlapping, so mosaics can be constructed from disjoint frames. On the other hand, for small-scale mosaics, where global deformations are relatively small, a registration system based on pattern matching and local distortion correction can be more accurate.
would you compare PI's registration ability...
With very few exceptions, I never comment on competing applications. And this is not one of those exceptions. We work very hard every day to design and implement the best tools we are capable of, with accuracy and versatility as our main aims. If this is not enough for somebody, then that's fine, use other options if they suit you better.
