Why the studios who are using RenderMan in pipeline such as Weta Digital {2} and Dreamworks {3} require an out-core ray tracer to process the massive point cloud ? Because in RenderMan, ambient occlusion pass and GI pass (inter-diffuse) are quite expensive, and at the same time, CPU-based solution such as ptfilter in RPS can't handle too huge point cloud or too slow.


This is not the rocket science, there are two critical keys about similar tool
* Out-of-core method to build accelerating structure from point cloud.
* Spherical harmonics-based AO and GI, as {3}, or simple approximated way.

Our plan

In a nutshell, no REYES pipeline, no requirement about this.
Milestones are,
  1. Milestone 1
    1. Frameworks for development.
    2. API design.
  2. Milestone 2
    1. Out-of-core mechanisms for accelerating structure.
  3. Milestone 3
    1. AO.
  4. Milestone 4
    1. GI
  5. Integration testing


{1} Point-Based Approximate Color Bleeding
{2} PantaRay: Fast Ray-traced Occlusion Caching of Massive Scenes
{3} Coherent Out-of-Core Point-Based Global Illumination

Last edited Dec 14, 2011 at 7:59 AM by JediMaster, version 1


No comments yet.