I'm currently working on a view space representation to accelerate an approximate ray-tracing technique on GPU. I would like to hear of similar attempts to compare some ideas. I'm using an A-Buffer like data structure, with some additional information for the efficient detection of ray-surface intersection from a view space perspective, which allows me to do an adaptive step sampling of the ray and find approximate surface intersections. I'm having a lot of troubles with under and over-sampling of the ray steps, hence missing surfaces, or spending too much time on empty spaces. I'd like to hear from any idea that offers a solution to similar problems. I'm interested in pure-GPU implementations.
- Badges
- Science topic
- Similar topics
- Papers
More Alejandro Piad Morffis's questions See All
Similar questions and discussions