Vladimir A. Galaktionov

Grid-based SAH BVH construction on a GPU

We present an efficient algorithm for building an adaptive bounding volume hierarchy (BVH) in linear time on commodity graphics hardware using CUDA. BVHs are widely used as an acceleration data structure to quickly ray trace animated polygonal scenes. We accelerate the construction process with auxiliary grids that help us build high quality BVHs with SAH in O(k∗n). We partition scene triangles and build a temporary grid structure only once. We also handle non-uniformly tessellated and long/thin triangles that we split into several triangle references with tight bounding box approximations. We make no assumptions on the type of geometry or animation motion. However, our algorithm takes advantage of coherent geometry layout and coherent frame-by-frame motion. We demonstrate the performance and quality of resulting BVHs that are built quickly with good spatial partitioning.

Analysis of the development concepts and methods of visual data representation in computational physics

Main steps in the development of scientific visualization as a branch of science are discussed. The evolution and prospects of the development of concepts, methods, and approaches of visual representation of numerical results obtained in computational physics (mainly, in computational fluid dynamics) are discussed.

Bidirectional Ray Tracing for the Integration of Illumination by the Quasi-Monte Carlo Method

Algorithms used to generate physically accurate images are usually based on the Monte Carlo methods for the forward and backward ray tracing. These methods are used to numerically solve the light energy transport equation (the rendering equation). Stochastic methods are used because the integration is performed in a high-dimensional space, and the convergence rate of the Monte Carlo methods is independent of the dimension. Nevertheless, modern studies are focused on quasi-random samples that depend on the dimension of the integration space and make it possible to achieve, under certain conditions, a high rate of convergence, which is necessary for interactive applications. In this paper, an approach to the development of an algorithm for the bidirectional ray tracing is suggested that reduces the overheads of the quasi-Monte Carlo integration caused by the high effective dimension and discontinuity of the integrand in the rendering equation. The pseudorandom and quasi-random integration methods are compared using the rendering equations that have analytical solutions.

An Effective Tone Mapping Operator for High Dynamic Range Images

Tone mapping operators are used to compress a large range of pixel luminances into a smaller range, which can be displayed on a monitor screen. In this paper, an effective and easy-to-use tone mapping operator based on the latest ideas developed in this field is presented. The parameter estimation process relies on the sampling method. Special attention is given to the robustness of the algorithm for the parameter estimation. The suggested tone mapping operator ensures good quality of images and almost does not require manual parameter tuning.

PARAMETRIC OPTIMIZING ANALYSIS OF UNSTEADY STRUCTURES AND VISUALIZATION OF MULTIDIMENSIONAL DATA

The paper presents an approach to fast approximate estimation of conditions for space-time structures appearing in the flows. The approach is based on combination of optimization problem computation with methods of data visual presentation. The visual presentation methods are applied for analysis of multidimensional array containing discrete result data. Optimization problem solution is implemented by parallel computation in a multitask form. For some cases, the approach allows to obtain for control parameter of considered problem the sought-for approximate dependence on characteristic parameters in a quasi-analytical form.

Spectral ray tracing in problems of photorealistic imagery construction

This paper analyzes color rendering problems arising in computer imaging of scenes with complex optical properties. The sources of these errors for imagery constructed on the basis of the RGB space are analyzed and an efficient and correct algorithm is proposed for spectral imagery. These solutions are applied to problems of direct stochastic and inverse deterministic ray tracing. In addition, complex scenes with mixed optical properties defined in different (including RGB) models are proposed to be treated by an efficient and correct algorithm for reducing all optical properties into a single spectral model.

Photorealistic volume scattering model in the bidirectional stochastic ray tracing problem

This paper is devoted to the development of physically correct rendering model of scenes containing volume scattering objects. The solution of the rendering equation is based on the Monte-Carlo bidirectional ray tracing. Two efficient approaches to the solution of the rendering equation for different parameters of the volume scattering medium (typically, concentration of scattering particles) are developed. Examples illustrate how the proposed models can be used for photorealistic visualization of scenes containing volume scattering objects and for the simulation of illuminators based on the volume scattering effect. Possible limitations of the application field of the proposed models are considered.

Multidimensional data analysis and visualization for time-dependent CFD problems

This paper is devoted to analysis and visualization of multidimensional data in problems of computational fluid dynamics (CFD). Multidimensional data are considered as the result of parametrical search and optimizing analysis. For such types of multidimensional data, analysis is aimed to find some hidden dependencies of specific parameters. This is the main difference of the proposed approach from the general approach of data analysis, which is usually applied to solve problems of classification and clusterization. A rough approximate approach is proposed for data processing. This approach includes data visualization in the space of principal components and data approximation by geometrical primitives (in particular, planes). An example of practical application of the proposed approach is given. The possibility of elastic maps application to multidimensional data in question is discussed.

Implementing irradiance cache in a GPU realistic renderer

This work presents an approach to integrating irradiance caching (IC) technique in a complete GPU photorealistic renderer. This work proposes a GPU friendly IC solution, where performance critical parts of an irradiance cache algorithm are done completely on the GPU. The modified algorithm for the GPU is different from a traditional implementation in 2 ways. The first distinction is a predictive nature of our algorithm that allows us to insert a large record set at once instead of inserting records one by one, as in traditional approaches. The second distinction is a new heuristic for validity radius computations. We also consider some low-level details and provide performance analysis of our solution.

Out-of-core GPU ray tracing of complex scenes

Increased demand for global illumination, image based-lighting and simplified workflow have pushed raytracing into mainstream. Many rendering and simulation algorithms that were considered strictly offline are becoming more interactive on massively parallel GPUs. Unfortunately, the amount of available memory on modern GPUs is relatively small. Scenes for feature film rendering and visualization have large geometric complexity and can easily contain millions of polygons and a large number of texture maps and other data attributes. In this talk, we describe a general purpose out-of-core ray tracing engine for the GPU where we address data management, ray-intersection and shading. We utilize a GPU data cache that enables efficient access of out-of-core data. We develop a novel ray intersection algorithm built around acceleration structure that brings needed data on demand using page-swapping. We further reduce memory usage by using a simple geometry quantization. The ray tracing engine is used to implement a variety of rendering and light transport algorithms.