Mickaël Ribardière

Spherical Fibonacci Point Sets for Illumination Integrals

Quasi‐Monte Carlo (QMC) methods exhibit a faster convergence rate than that of classic Monte Carlo methods. This feature has made QMC prevalent in image synthesis, where it is frequently used for approximating the value of spherical integrals (e.g. illumination integral). The common approach for generating QMC sampling patterns for spherical integration is to resort to unit square low‐discrepancy sequences and map them to the hemisphere. However such an approach is suboptimal as these sequences do not account for the spherical topology and their discrepancy properties on the unit square are impaired by the spherical projection. In this paper we present a strategy for producing high‐quality QMC sampling patterns for spherical integration by resorting to spherical Fibonacci point sets. We show that these patterns, when applied to illumination integrals, are very simple to generate and consistently outperform existing approaches, both in terms of root mean square error (RMSE) and image quality. Furthermore, only a single pattern is required to produce an image, thanks to a scrambling scheme performed directly in the spherical domain.

A Spherical Gaussian Framework for Bayesian Monte Carlo Rendering of Glossy Surfaces

The Monte Carlo method has proved to be very powerful to cope with global illumination problems but it remains costly in terms of sampling operations. In various applications, previous work has shown that Bayesian Monte Carlo can significantly outperform importance sampling Monte Carlo thanks to a more effective use of the prior knowledge and of the information brought by the samples set. These good results have been confirmed in the context of global illumination but strictly limited to the perfect diffuse case. Our main goal in this paper is to propose a more general Bayesian Monte Carlo solution that allows dealing with nondiffuse BRDFs thanks to a spherical Gaussian-based framework. We also propose a fast hyperparameters determination method that avoids learning the hyperparameters for each BRDF. These contributions represent two major steps toward generalizing Bayesian Monte Carlo for global illumination rendering. We show that we achieve substantial quality improvements over importance sampling at comparable computational cost.

Improving Performance and Accuracy of Local PCA

Local Principal Component Analysis (LPCA) is one of the popular techniques for dimensionality reduction and data compression of large data sets encountered in computer graphics. The LPCA algorithm is a variant of k‐means clustering where the repetitive classification of high dimensional data points to their nearest cluster leads to long execution times. The focus of this paper is on improving the efficiency and accuracy of LPCA. We propose a novel SortCluster LPCA algorithm that significantly reduces the cost of the point‐cluster classification stage, achieving a speed‐up of up to 20. To improve the approximation accuracy, we investigate different initialization schemes for LPCA and find that the k‐means++ algorithm [ AV07 ] yields best results, however at a high computation cost. We show that similar ideas that lead to the efficiency of our SortCluster LPCA algorithm can be used to accelerate k‐means++. The resulting initialization algorithm is faster than purely random seeding while producing substantially more accurate data approximation.

Adaptive Records for Irradiance Caching

Irradiance Caching is one of the most widely used algorithms to speed up global illumination. In this paper, we propose an algorithm based on the Irradiance Caching scheme that allows us (1) to adjust the density of cached records according to illumination changes and (2) to efficiently render the high‐frequency illumination changes. To achieve this, a new record footprint is presented. Although the original method uses records having circular footprints depending only on geometrical features, our record footprints have a more complex shape which accounts for both geometry and irradiance variations. Irradiance values are computed using a classical Monte Carlo ray tracing method that simplifies the determination of nearby objects and the pre‐computation of the shape of the influence zone of the current record. By gathering irradiance due to all the incident rays, illumination changes are evaluated to adjust the footprint’s records. As a consequence, the record footprints are smaller where illumination gradients are high. With this technique, the record density depends on the irradiance variations. Strong variations of irradiance (due to direct contributions for example) can be handled and evaluated accurately. Caching direct illumination is of high importance, especially in the case of scenes having many light sources with complex geometry as well as surfaces exposed to daylight. Recomputing direct illumination for the whole image can be very time‐consuming, especially for walkthrough animation rendering or for high‐resolution pictures. Storing such contributions in the irradiance cache seems to be an appropriate solution to accelerate the final rendering pass.

STD: Student's t-Distribution of Slopes for Microfacet Based BSDFs

This paper focuses on microfacet reflectance models, and more precisely on the definition of a new and more general distribution function, which includes both Beckmanns and GGX distributions widely used in the computer graphics community. Therefore, our model makes use of an additional parameter γ, which controls the distribution function slope and tail height. It actually corresponds to a bivariate Students t‐distribution in slopes space and it is presented with the associated analytical formulation of the geometric attenuation factor derived from Smith representation. We also provide the analytical derivations for importance sampling isotropic and anisotropic materials. As shown in the results, this new representation offers a finer control of a wide range of materials, while extending the capabilities of fitting parameters with captured data.

Adaptive records for volume irradiance caching

In this paper, we present a new irradiance caching scheme using Monte Carlo ray tracing for efficiently rendering participating media. The irradiance cache algorithm is extended to participating media. Our method allows to adjust the density of cached records depending on illumination changes. Direct and indirect contributions can be stored in the records but also multiple scattering. An adaptive shape of the influence zone of records, depending on geometrical features and irradiance variations, is introduced. To avoid a high density of cached records in low interest areas, a new method controls the density of the cache when adding new records. This record density control depends on the interpolation quality and on the photometric characteristics of the medium. Reducing the number of records accelerates both the computation pass and the rendering pass by decreasing the number of queries to the cache data structure (Kd-tree). Finally, instead of using an expensive ray marching to find records that cover the ray, we gather all the contributive records along the ray. With our method, pre-computing and rendering passes are significantly speeded-up.

Eye-Centered Color Adaptation in Global Illumination

Color adaptation is a well known ability of the human visual system (HVS). Colors are perceived as constant even though the illuminant color changes. Indeed, the perceived color of a diffuse white sheet of paper is still white even though it is illuminated by a single orange tungsten light, whereas it is orange from a physical point of view. Unfortunately global illumination algorithms only focus on the physics aspects of light transport. The ouput of a global illuminantion engine is an image which has to undergo chromatic adaptation to recover the color as perceived by the HVS. In this paper, we propose a new color adaptation method well suited to global illumination. This method estimates the adaptation color by averaging the irradiance color arriving at the eye. Unlike other existing methods, our approach is not limited to the view frustrum, as it considers the illumination from all the scene. Experiments have shown that our method outperforms the state of the art methods.

Rendering Rough Opaque Materials with Interfaced Lambertian Microfacets

Specular microfacet distributions have been successfully employed by many authors for representing glossiness of materials. They are generally combined with a Lambertian term to account for the colored aspect. These representations make use of the Fresnel reflectance factor at the interface, but the transmission factor at the interface should also be managed. One solution is to employ a multi-layered model with a single layer for the rough interface, which requires a numerical simulation for handling the multiple reflections of light between the substrate and the interface. In this paper, we propose rather to use a representation corresponding to a Fresnel interface lying on a Lambertian substrate, for which the multiple reflections of light between the interface and the substrate can be expressed analytically. With this interfaced Lambertian model, we show how Fresnel transmission affects the material appearance for flat and rough surfaces with isotropic and anisotropic distributions, that produce light backscattering effects. We also propose a methodology for using such materials in any physically based Monte Carlo rendering system, as well as an approximate representation, suitable for GPU applications or measured data fitting. Our approach generalizes several previous models, including flat Lambertian materials as well as specular and Lambertian microfacets. Our results illustrate the wide range of materials that can be rendered with this representation.

Appearance of Interfaced Lambertian Microfacets, using STD Distribution

This paper presents the use of Students T-Distribution (STD) with interfaced Lambertian (IL) microfacets. The resulting model increases the range of materials while providing a very accurate adjustment of appearance. STD has been recently proposed as a generalized distribution of microfacets which includes Beckmann and GGX widely used in computer graphics; IL corresponds to a physical representation of a Lambertian substrate covered with a flat Fresnel interface. We illustrate the appearance variations that can be observed, and discuss the advantages of using such a combination.

A radiance cache method for highly glossy surfaces

Radiance caching methods have proven to be efficient for global illumination. Their goal is to compute precisely illumination values (incident radiance or irradiance) at a reasonable number of points lying on the scene surfaces. These points, called records, are stored in a cache used for estimating illumination at other points in the scene. Unfortunately, with records lying on glossy surfaces, the irradiance value alone is not sufficient to evaluate the reflected radiance; each record should also store the incident radiance for all incident directions. Memory storage can be reduced with projection techniques using spherical harmonics or other basis functions. These techniques provide good results for low shininess BRDFs. However, they get impractical for shininess of even moderate value, since the number of projection coefficients increases drastically. In this paper, we propose a new radiance caching method that handles highly glossy surfaces while requiring a low memory storage. Each cache record stores a coarse representation of the incident illumination thanks to a new data structure, called Equivalent Area light Sources, capable of handling fuzzy mirror surfaces. In addition, our method proposes a new simplification of the interpolation process, since it avoids the need for expressing and evaluating complex gradients.