Wyniki wyszukiwania - BazTech

1

Hierarchical stochastic radiosity for implicitly defined surfaces

Tobler R. F., Purgathofer W.

Machine Graphics and Vision

|

2000

|

Vol. 9, No. 3

561-571

EN

The method presented in this paper calculates global illumination for scenes with implicitly defined surfaces, such as CSG models. A Monte Carlo type particle tracer which directly uses the CSG models distributes the light power in the scene. Global illumination information can be maintained at different resolution levels with the help of a hierarchical storage mechanism for radiosity approximations as the simulation of aprticle proceeds. Shadow leaks and excessive variance of the radiosity values are avoided by filtering the illumination information represented in the octree data structure. Included results show that new method is viable for generating high quality global illumination solutions.

2

The constant radiance term

Neumann L., Neumann A., Prikryl J., Purgathofer W.

Machine Graphics and Vision

|

1998

|

Vol. 7, No. 3

535-549

EN

This paper describes a fundamential improvement for all rendering methods that evaluate the rendering equation in any complexity. The "constant Radiance Term" accelerates the computation of multiple interreflections in arbitrary non-diffuse environments. It is an extension of the constant radiosity step untroduced ealiner for solving radiosity problems and does not require geometrical nor visibility information. A constant radiance portion is extracted from the solution in every surface point and every direction. The residuum problem can be solved faster or with reduced variance by all known methods that consider non-diffuse light interreflections, like different versions of distribution ray tracing, deterministic and stochastic radiosity methods, and hybrid solutions. The constant radiance value is chosen in such a way that the total self-emitted power of the residuum problem is zero so that after the extraction of the constant part there are places with positive self-emittance and places with positive self-emittance and places with negative self emittance. The achieved acceleration can be significant for environments with many bright coloured surfaces. Using a decomposition into separate positive and negative problems it is possible to apply the technique even when using turnkey rendering software.

3

Non-diffuse, random-walk radiosity algorithm with linear basis functions

Szirmay-Kalos L., Foris T., Purgathofer W.

Machine Graphics and Vision

|

1998

|

Vol. 7, No. 1/2

475-484

EN

This paper presents an efficient method to solve the general rendering equation, using a combined finite element and quasi-random walk approach. Applying point collocation method, the surfaces are decomosed into planar patches where the directional distribution of the radiance is assumed to be a linear combination of the disrtibutions at the vertices. The direction dependet radiance function of the vertces is then computed by random of quasi-random walk.