Core: better shape prediction algorithm
Variation in spot shape is purely an effect of detector geometry. This is why for a cylindrical detector like BioDiff, the spots at the top and bottom of the detector image are the most distorted and difficult to integrate. In reciprocal space, the reflection shapes are independent of position.
Therefore, there are a couple of possibilities that would allow us to average over all strong peaks to get a guess for the shape, rather than over local peaks.
-
Transform the detector image from a cylinder to a sphere, then integrate as usual.
-
Transform the detector spots to 3D reciprocal space ellipsoids. Compute a mean ellipsoid, then transform the resulting shape back to a detector position-dependent ellipse.