I am working on this right now. The problem is not in ImageCalibration, which is fully parallelized at a high level (if you feed it with enough images, it should saturate all of your processor cores). The main performance issues are in StarAlignment and ImageIntegration. The next versions of these processes will be much faster, especially SA, which is currently the main preprocessing bottleneck. The performance improvements should be quite spectacular. I just ask for a bit of patience because the things I am working on are complex and delicate.
Besides being much faster for batch image registration tasks, StarAlignment will implement a completely new distortion correction algorithm. This will be an important improvement for mosaic construction and registration of wide field images acquired with different instruments. Patience
