Jump to content
  • Advertisement
Sign in to follow this  
gjaegy

Clustered Principal Component Analysis (CPCA) compression

This topic is 4859 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Hi, I've successfully implemented SH lighting, thanks to the great paper written by Green. However, I would like to implement some compression solution in order to reduce the ammount of datas needed for the SH coefficients. I think one of the best method is the one used by DirectX, developped by Sloan, called CPCA. Has anyone already implemented this solution? Does anyone know some papers describing the implementation (apart from the original one)? Thanks for your help, Greg

Share this post


Link to post
Share on other sites
Advertisement
I have heard a lecture about using CPCA for statistical pattern recognition 1 week ago.

It's not that complicated though pretty abstract:
1) divide you data into clusters
2) find a gaussian distribution for each cluster
3) compute the largest eigenvalue of the covariance matrix of each gaussian
4) find the eigenvector corresponding to the largest eigenvalue
5) represent your data as data(x)=centerofclusterof(x)+t(x)*eigenvector_of_cluster+errorterm(x)
6) the errorterm is usually pretty small and can be represented using a few bits.

Ah, wikipedia has some stuff too: wikipedia

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

GameDev.net is your game development community. Create an account for your GameDev Portfolio and participate in the largest developer community in the games industry.

Sign me up!