Vertex Cache Optimised Mesh Compression
This is a proof of concept library expanding on previously presented index buffer compression mechanisms and adding compression for vertices as well.
How does it work?
Basically, it expands on the index buffer compression using a parallelogram predictor and a kind of universal code (related to exponential golomb) along with an exponential moving average to give fast adaptive compression for vertex attributes. A more complete version of this is outlined in this blog post.