Automatic and Dynamic Generation of GPU Programs to Mix Renderings and Enhance Informations on High Scale Scenes

V. Boyer (France)


Rendering, GPU programming, Non-Photorealistic Com puter Graphics.


We propose an automatic and dynamic generation of GPU programs to mix renderings and enhance informa tions for visualization of high scale scenes. Mixing dif ferent kinds of rendering techniques in the same frame en hances pertinent informations on 2D images or 3D scenes. This is achieved in one pass because one and only one GPU program is used to render the scene. New rendering tech niques can be imported and we present a method to auto matically build and dynamically updated the program used by the GPU. For a given 3D scene or a 2D image, the user chooses key points, imports rendering techniques described as GPU programs and our model constructs one GPU pro gram that render the scene with these different rendering techniques. The key points, the rendering techniques used, the camera position can be interactively changed by the user. In this paper we present the model used to mix render ings, a method to automatically generate and dynamically update GPU programs. The images produced by our sys tem are presented and results are discussed.

Important Links:

Go Back