pattacini
commented
5 years ago Deformation could be a good addition to consider, although I would rather plain it as an upcoming development to avoid having too many ingredients to cook together.
Regarding the related gradient, it would be nice to have an exhaustive analysis.
Since a new structure of the code will be provided, it'd worth trying to include also deformations both in the single- and multiple-superquadric modeling. This would require to estimate some extra parameters (for a maximum of 15) during the superquadric fitting. Actually, due to the more complex expressions of the inside-outside function in this case, I think that only the finite-differences implementation will be duable.