@IgorCarron You got a thank you: http://t.co/4EWJxgc98y
— Robin Green (@fatlimey) March 20, 2014
I learned that Manny's slides for this year's Game Developer's Conference were out. Manny uses a specific matrix factorization (dictionary learning) for the decomposition of the cloth movement in animated movie characters (check Skinnned cloth). Back in 1995-96, Texas A&M had started the Student Research Week and I had been asked to be a judge or something. I specifically recall mentioning to Richard DeVaul (now at Google X) that his poster featuring a technique for speeding up the computations of the cloth/draping in visualization looked very much like an instance of the Fast Multipole Method (FMM). Fast forward to now, FMM is really an instance of matrix (kernel) factorization that takes a kernel that is a function of x and x' as a series of products of one function of x and another function of x' and does so with specific constraints in the decomposition (based on distance the expansion is different). Wow. Here are the slides:
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.
Post a Comment