Jean Claude Iehl

Non-interleaved deferred shading of interleaved sample patterns

This paper presents a novel and fast technique to combine interleaved sampling and deferred shading on a GPU. The core idea of this paper is quite simple. Interleaved sample patterns are computed in a non-interleaved deferred shading process. The geometric buffer (G-buffer) which contains all of the pixel information is actually split into several separate and distinct sub-buffers. To achieve such a result in a fast way, a massive two-pass swizzling copy is used to convert between these two buffer organizations. Once split, the sub-buffers can then be accessed to perform any fragment operation as it is done with a standard deferred shading rendering pipeline. By combining interleaved sampling and deferred shading, real time rendering of global illumination effects can be therefore easily achieved. Instead of evaluating each light contribution on the whole geometric buffer, each shading computation is coherently restricted to a smaller subset a fragments using the sub-buffers. Therefore, each screen pixel in a regular n X m pattern will have its own small set of light contributions. Doing so, the consumed fillrate is considerably decreased and the provided rendering quality remains close to the quality obtained with a non-interleaved approach. The implementation of this rendering pipeline is finally straightforward and it can be easily integrated in any existing real-time rendering package already using deferred shading.

Bidirectional instant radiosity

This paper presents a new sampling strategy to achieve interactive global illumination on one commodity computer. The goal is to propose an efficient numerical stochastic scheme which can be well adapted to a fast rendering algorithm. As we want to provide an efficient sampling strategy to handle difficult settings without sacrificing performance in common cases, we developed an extension of Instant Radiosity [Kel97] in the same way bidirectional path tracing is an extension of path or light tracing. Our idea is to build several estimators and to efficiently combine them to find a set of virtual point light sources which are relevant for the areas of the scene seen by the camera. The resulting algorithm is faster than classical solutions to global illumination. Using today graphics hardware, an interactive frame rate and the convergence of the scheme can be easily obtained in scenes with many light sources, glossy materials or difficult visibility problems.

Linear efficient antialiased displacement and reflectance mapping

We present Linear Efficient Antialiased Displacement and Reflectance (LEADR) mapping, a reflectance filtering technique for displacement mapped surfaces. Similarly to LEAN mapping, it employs two mipmapped texture maps, which store the first two moments of the displacement gradients. During rendering, the projection of this data over a pixel is used to compute a noncentered anisotropic Beckmann distribution using only simple, linear filtering operations. The distribution is then injected in a new, physically based, rough surface microfacet BRDF model, that includes masking and shadowing effects for both diffuse and specular reflection under directional, point, and environment lighting. Furthermore, our method is compatible with animation and deformation, making it extremely general and flexible. Combined with an adaptive meshing scheme, LEADR mapping provides the very first seamless and hardware-accelerated multi-resolution representation for surfaces. In order to demonstrate its effectiveness, we render highly detailed production models in real time on a commodity GPU, with quality matching supersampled ground-truth images.

Metropolis Instant Radiosity

We present Metropolis Instant Radiosity (MIR), an unbiased algorithm to solve the Light Transport problem. MIR is a hybrid technique which consists in representing the incoming radiance field by a set of Virtual Point Lights (V PLs) and in computing the response of all sensors in the scene (i.e. camera captors) by accumulating their contributions. In contrast to other similar approaches, we propose to sample the VPLs with an innovative Multiple‐try Metropolis‐Hastings (MTMH) Algorithm: the goal is to build an efficient, aggressive, and unconditionally robust variance reduction method that works well regardless of the scene layout. Finally, we present a fast ray tracing implementation using MIR and show how our complete rendering pipeline can produce high‐quality and high‐resolution pictures in a few seconds.

An Adaptive Spectral Rendering with a Perceptual Control

In this paper, we present a spectral rendering method based on a ray tracing algorithm and guided by a perceptual control of the error made. An adaptive representation of spectral data for light sources and materials is used and induces, for each pixel, the evaluation of an algebraic expression. For each visible wavelength, the computations of complex spread in refractive and dispersive materials, based on several photon maps, are locally restricted by an adaptive evaluation of the expression. This method allows to simulate high quality physically‐based pictures and, in particular, some specific phenomena like dispersion in transparent objects, scattered caustics,. . .

Extracting Microfacet-based BRDF Parameters from Arbitrary Materials with Power Iterations

We introduce a novel fitting procedure that takes as input an arbitrary material, possibly anisotropic, and automatically converts it to a microfacet BRDF. Our algorithm is based on the property that the distribution of microfacets may be retrieved by solving an eigenvector problem that is built solely from backscattering samples. We show that the eigenvector associated to the largest eigenvalue is always the only solution to this problem, and compute it using the power iteration method. This approach is straightforward to implement, much faster to compute, and considerably more robust than solutions based on nonlinear optimizations. In addition, we provide simple conversion procedures of our fits into both Beckmann and GGX roughness parameters, and discuss the advantages of microfacet slope space to make our fits editable. We apply our method to measured materials from two large databases that include anisotropic materials, and demonstrate the benefits of spatially varying roughness on texture mapped geometric models.

Towards perceptual control of physically based spectral rendering

Abstract Realistic rendering research focuses on generating images that should be perceived the same way as a real scene. We will describe a physically based spectral framework where energy, materials and light field are wavelength dependent. To reduce the computational load we also make use of the high variations of the human vision sensitivity to adapt the accuracy of every computation. As no comprehensive human vision model exists, we use several types of perceptual control. In order to adapt computation accuracy we have also derived a simple but useful perceptual error estimate. This paper will introduce some ideas.

Improving robustness of Monte-Carlo global illumination with directional regularization

Directional regularization offers great potential to improve the convergence rates of Monte-Carlo-based global illumination algorithms. In this paper, we show how it can be applied successfully by combining unbiased bidirectional strategies, photon mapping, and biased directional regularization.

Interactive Curvature Tensor Visualization on Digital Surfaces

Interactive visualization is a very convenient tool to explore complex scientific data or to try different parameter settings for a given processing algorithm. In this article, we present a tool to efficiently analyze the curvature tensor on the boundary of potentially large and dynamic digital objects mean and Gaussian curvatures, principal curvatures, principal directions and normal vector field. More precisely, we combine a fully parallel pipeline on GPU to extract an adaptive triangulated isosurface of the digital object, with a curvature tensor estimation at each surface point based on integral invariants. Integral invariants being parametrized by a given ball radius, our proposal allows to explore interactively different radii and thus select the appropriate scale at which the computation is performed and visualized.

High performance ultrasonic field simulation on complex geometries

Ultrasonic field simulation is a key ingredient for the design of new testing methods as well as a crucial step for NDT inspection simulation. As presented in a previous paper [1], CEA-LIST has worked on the acceleration of these simulations focusing on simple geometries (planar interfaces, isotropic materials). In this context, significant accelerations were achieved on multicore processors and GPUs (Graphics Processing Units), bringing the execution time of realistic computations in the 0.1 s range.In this paper, we present recent works that aim at similar performances on a wider range of configurations. We adapted the physical model used by the CIVA platform to design and implement a new algorithm providing a fast ultrasonic field simulation that yields nearly interactive results for complex cases. The improvements over the CIVA pencil-tracing method include adaptive strategies for pencil subdivisions to achieve a good refinement of the sensor geometry while keeping a reasonable number of ray-tracing o...