Steep parallax mapping
Author
I implemented parallax mapping but I'm not happy with the result, is not very different from normal mapping (unless I made some mistake).
I was reading about possible improvements, and found about Steep Parallax Mapping.
I cant find any demo that implements it on the net, I really need to know the performance compared to normal parallax.
So I tried to implement it, and I found this website
http://graphics.cs.brown.edu/games/SteepParallax/index.html
It contains fragment shader code, but not vertex shader, I wonder what those two mean
varying vec3 tsL;
varying vec3 _tsE;
I presume ts means tangent space, so tsL would be light vector in tangent space, but what about tsE?
Quote:Original post by Relfos
I presume ts means tangent space, so tsL would be light vector in tangent space, but what about tsE?
Eye ?
Just as a public service announcement, quadtree/image pyramid-based methods are more in line with current state-of-the-art.
Re: performance-- It's somewhat slower, depending on the hardware. If anything I believe techniques like the above might actually run slightly faster if you're clever about explicit LOD samples, as you can factor out/amortize the rather expensive derivative calculations for texture filtering. You're obviously doing more texture samples, though, which is a prime candidate for bottlenecking.
EDIT: Munged up my link. Fixed now.
Re: performance-- It's somewhat slower, depending on the hardware. If anything I believe techniques like the above might actually run slightly faster if you're clever about explicit LOD samples, as you can factor out/amortize the rather expensive derivative calculations for texture filtering. You're obviously doing more texture samples, though, which is a prime candidate for bottlenecking.
EDIT: Munged up my link. Fixed now.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
