Gael Sourimant

GPU Shape Grammars

GPU Shape Grammars provide a solution for interactive procedural generation, tuning and visualization of massive environment elements for both video games and production rendering. Our technique generates detailed models without explicit geometry storage. To this end we reformulate the grammar expansion for generation of detailed models at the tesselation control and geometry shader stages. Using the geometry generation capabilities of modern graphics hardware, our technique generated massive, highly detailed models. GPU Shape Grammars integrate within a scalable framework by introducing automatic generation of levels of detail at reduced cost. We apply our solution for interactive generation and rendering of scenes containing thousands of buildings and trees.

Sub-pixel shadow mapping

The limited resolution of shadow maps may result in erroneous shadowing, yielding artificially jagged edges (Figure 1) and temporally crawling shadows even using perspective optimization techniques. Dai et al. [2008] propose an explicit storage of geometry within shadow map texels to avoid aliasing. Each texel stores the coordinates of the closest triangle only, potentially leading to false negatives in the intersection computation while incurring large memory consumption. These artifacts are reduced by intersecting the triangles stored in numerous neighboring texels, resulting in significant performance hit while still missing some intersections. We introduce Sub-Pixel Shadow Maps (SPSM) for real-time shadow mapping with sub-pixel precision. Our technique is based on the storage of a fixed-size partial representation of the scene geometry using conservative rasterization, combined with an original reconstruction of shadow edges.

Accurate analytic approximations for real-time specular area lighting

Accurate real-time rendering of specular surfaces is a challenging task when considering area light source illumination. The difficulty resides in the evaluation of a surface integral for which no practical solution exists except using expensive Monte Carlo methods. Recent techniques like the most representative point approach [Drobot 2014] alleviate this problem but make some accuracy trade-off to achieve real-time performance. We introduce analytic approximations for accurate real-time rendering of specular surfaces illuminated by polygonal light sources. Our solution is based on a reformulation of the contour integral [Arvo 1995] we approximate analytically with simple peak functions. In addition, using simple geometric operations, we extend the solution to handle more physically plausible BRDFs. Our solution works without any assumption on light source shape nor surface roughness, bringing real-time performances with a quality close to the ground truth.

Poisson disk ray-marched ambient occlusion

Ambient occlusion (AO) [Pharr and Green 2004] provides an approximation to global illumination. It can be computed in real-time if restricted to screen-space [Mittring 2007; Loos and Sloan 2010], at the cost of an approximation of solutions computed in geometric space, thus introducing a loss of quality.

Analytic Approximations for Real-Time Area Light Shading

We introduce analytic approximations for accurate real-time rendering of surfaces lit by non-occluded area light sources. Our solution leverages the Irradiance Tensors developed by Arvo for the shading of Phong surfaces lit by a polygonal light source. Using a reformulation of the 1D boundary edge integral, we develop a general framework for approximating and evaluating the integral in constant time using simple peak shape functions. To overcome the Phong restriction, we propose a low cost edge splitting strategy that accounts for the spherical warp introduced by the half vector parametrization. Thanks to this novel extension, we accurately approximate common microfacet BRDFs, providing a practical method producing specular stretches that closely match the ground truth in real-time. Finally, using the same approximation framework, we introduce support for spherical and disc area light sources, based on an original polygon spinning method supporting non-uniform scaling operations and horizon clipping. Implemented on a GPU, our method achieves real-time performances without any assumption on area light shape nor surface roughness.

Triple depth culling

Virtual worlds feature increasing geometric and shading complexities, resulting in a constant need for effective solutions to avoid rendering objects invisible for the viewer. This observation is particularly true in the context of real-time rendering of highly occluded environments such as urban areas, landscapes or indoor scenes. This problem has been intensively researched in the past decades, resulting in numerous optimizations building upon the well-known Z-buffer technique. Among them, extensions of graphics hardware such as early Z-culling [Morein 2000] efficiently avoid shading most of invisible fragments. However, this technique is not applicable when the fragment shader discards fragments or modifies their depth value, or if alpha testing is enabled [nVidia 2008].