I currently use SVD from numpy / scipy for polar decomposition as part of computing least-squares rigid body transformations in biomechanics research (using the algorithm from [1]). Blender is the best way I currently have to visualise the results. I can imagine many similar applications in tracking, motion capture, etc.
I use the linear algebra capabilities from numpy quite a bit though, so access to routines for arbitrary-sized matrix operations, easy matrix assembly, determinant, trace, and linear system solvers would also be necessary in order for me not to import numpy. Not that my use of Blender is typical. :-) Jonathan Merritt. [1] - Veldpaus, F.E., Woltring, H.J. and Dortmans, L.J.M.G. (1988) A least-squares algorithm for the equiform transformation from spatial marker coordinates. Journal of Biomechanics, 21(1):45-54. On Mon, Apr 23, 2012 at 8:05 AM, <mgsch...@blensor.org> wrote: > Hi Howard, > > the SVD decomposition could easily be added to the Eigen3 module (the > patch I mentioned in the first post), > but currently it seems we are agreeing on numpy integration anyway. > > Michael > > Am 23.04.2012 13:58, schrieb Howard Trickey: > > I too recently felt the need for more numerical stuff in the py api. > > In particular, I would have liked a Singular Value Decomposition > > for a best-fit-points-to-plane function. numpy has such a thing > > but no doubt the existing C/C++ source has it somewhere, maybe > > multiple times, already. > > > > - Howard > > _______________________________________________ > > Bf-committers mailing list > > Bf-committers@blender.org > > http://lists.blender.org/mailman/listinfo/bf-committers > > _______________________________________________ > Bf-committers mailing list > Bf-committers@blender.org > http://lists.blender.org/mailman/listinfo/bf-committers > _______________________________________________ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers