I can add my multivariate cubic interpolation because it is really fast.
BUT that includes the code to take a small section of the matrix to work with.
SO I can also extend that code to be the bspline optimisation on a smaller secton.
For the multivariate method I take a secton of cube the size I need - eh 8 points for a linear, 4x4 for a cubic.
For the optimised b-spline I have both degree and cube size seperately but it is the same idea. The m ain thing is getting the right size cube, and then when interpolating adjustuinfg the point into the right place in the cube. That code is already there my multivariate interpolation so extend into it for a cleaner method.
So optimised thevenax is a new kind of interpolator - leave the "entire cube" as an option. One inherited from the other. Which way not sure, but the OptimisedThevenaz is a kind of both bspline and multivariate.
I can add my multivariate cubic interpolation because it is really fast. BUT that includes the code to take a small section of the matrix to work with. SO I can also extend that code to be the bspline optimisation on a smaller secton.
For the multivariate method I take a secton of cube the size I need - eh 8 points for a linear, 4x4 for a cubic. For the optimised b-spline I have both degree and cube size seperately but it is the same idea. The m ain thing is getting the right size cube, and then when interpolating adjustuinfg the point into the right place in the cube. That code is already there my multivariate interpolation so extend into it for a cleaner method.
So optimised thevenax is a new kind of interpolator - leave the "entire cube" as an option. One inherited from the other. Which way not sure, but the OptimisedThevenaz is a kind of both bspline and multivariate.