Abstract Physically based rendering is a well‐understood technique to produce realistic‐looking images. However, different algorithms exist for efficiency reasons, which work well in certain cases but fail or produce rendering artefacts in others. Few tools allow a user to gain insight into the algorithmic processes. In this work, we present such a tool, which combines techniques from information visualization and visual analytics with physically based rendering. It consists of an interactive parallel coordinates plot, with a built‐in sampling‐based data reduction technique to visualize the attributes associated with each light sample. Two‐dimensional (2D) and three‐dimensional (3D) heat maps depict any desired property of the rendering process. An interactively rendered 3D view of the scene displays animated light paths based on the user's selection to gain further insight into the rendering process. The provided interactivity enables the user to guide the rendering process for more efficiency. To show its usefulness, we present several applications based on our tool. This includes differential light transport visualization to optimize light setup in a scene, finding the causes of and resolving rendering artefacts, such as fireflies, as well as a path length contribution histogram to evaluate the efficiency of different Monte Carlo estimators.

We present a novel method to optimize the attenuation of light for the single scattering model in direct volume rendering. A common problem of single scattering is the high dynamic range between lit and shadowed regions due to the exponential attenuation of light along a ray. Moreover, light is often attenuated too strong between a sample point and the camera, hampering the visibility of important features. Our algorithm employs an importance function to selectively illuminate important structures and make them visible from the camera. With the importance function, more light can be transmitted to the features of interest, while contextual structures cast shadows which provide visual cues for perception of depth. At the same time, more scattered light is transmitted from the sample point to the camera to improve the primary visibility of important features. We formulate a minimization problem that automatically determines the extinction along a view or shadow ray to obtain a good balance between sufficient transmittance and attenuation. In contrast to previous approaches, we do not require a computationally expensive solution of a global optimization, but instead provide a closed-form solution for each sampled extinction value along a view or shadow ray and thus achieve interactive performance.

We present a novel method to compute anisotropic shading for direct volume rendering to improve the perception of the orientation and shape of surface-like structures. We determine the scale-aware anisotropy of a shading point by analyzing its ambient region. We sample adjacent points with similar scalar values to perform a principal component analysis by computing the eigenvectors and eigenvalues of the covariance matrix. In particular, we estimate the tangent directions, which serve as the tangent frame for anisotropic bidirectional reflectance distribution functions. Moreover, we exploit the ratio of the eigenvalues to measure the magnitude of the anisotropy at each shading point. Altogether, this allows us to model a data-driven, smooth transition from isotropic to strongly anisotropic volume shading. In this way, the shape of volumetric features can be enhanced significantly by aligning specular highlights along the principal direction of anisotropy. Our algorithm is independent of the transfer function, which allows us to compute all shading parameters once and store them with the data set. We integrated our method in a GPU-based volume renderer, which offers interactive control of the transfer function, light source positions, and viewpoint. Our results demonstrate the benefit of anisotropic shading for visualization to achieve data-driven local illumination for improved perception compared to isotropic shading.

Abstract Inspired by vector field topology, an established tool for the extraction and identification of important features of flows and vector fields, we develop means for the analysis of the structure of light transport. For that, we derive an analogy to vector field topology that defines coherent structures in light transport. We also introduceFinite‐Time Path Deflection(FTPD), a scalar quantity that represents the deflection characteristic of all light transport paths passing through a given point in space. For virtual scenes, the FTPD can be computed directly using path‐space Monte Carlo integration. We visualize the FTPD field for several example scenes and discuss the revealed structures. Lastly, we show that the coherent regions visualized by the FTPD are closely related to the coherent regions in our new topologically‐motivated analysis of light transport. FTPD visualizations are thus also visualizations of the structure of light transport.

We introduce a refractive radiative transfer equation to the graphics community for the physically based rendering of participating media that have a spatially varying index of refraction. We review principles of geometric nonlinear optics that are crucial to discuss a more generic light transport equation. In particular, we present an optical model that has an integral form suitable for rendering. We show rigorously that the continuous bending of light rays leads to a nonlinear scaling of radiance. To obtain physically correct results, we build on the concept of basic radiance—known from discontinuous refraction—to conserve energy in such complex media. Furthermore, the generic model accounts for the reduction in the speed of light due to the index of refraction to render transient effects like the propagation of light echoes. We solve the refractive volume rendering equation by extending photon mapping with transient light transport in a refractive, participating medium. We demonstrate the impact of our approach on the correctness of rendered images of media that are dominated by spatially continuous refraction and multiple scattering. Furthermore, our model enables us to render visual effects like the propagation of light echoes or time-of-flight imagery that cannot be produced with previous approaches.

We present a novel and efficient method to compute volumetric soft shadows for interactive direct volume visualization to improve the perception of spatial depth. By direct control of the softness of volumetric shadows, disturbing visual patterns due to hard shadows can be avoided and users can adapt the illumination to their personal and application-specific requirements. We compute the shadowing of a point in the data set by employing spatial filtering of the optical depth over a finite area patch pointing toward each light source. Conceptually, the area patch spans a volumetric region that is sampled with shadow rays; afterward, the resulting optical depth values are convolved with a low-pass filter on the patch. In the numerical computation, however, to avoid expensive shadow ray marching, we show how to align and set up summed area tables for both directional and point light sources. Once computed, the summed area tables enable efficient evaluation of soft shadows for each point in constant time without shadow ray marching and the softness of the shadows can be controlled interactively. We integrated our method in a GPU-based volume renderer with ray casting from the camera, which offers interactive control of the transfer function, light source positions, and viewpoint, for both static and time-dependent data sets. Our results demonstrate the benefit of soft shadows for visualization to achieve user-controlled illumination with many-point lighting setups for improved perception combined with high rendering speed.

We present ambient scattering as a preintegration method for scattering on mesoscopic scales in direct volume rendering. Far-range scattering effects usually provide negligible contributions to a given location due to the exponential attenuation with increasing distance. This motivates our approach to preintegrating multiple scattering within a finite spherical region around any given sample point. To this end, we solve the full light transport with a Monte-Carlo simulation within a set of spherical regions, where each region may have different material parameters regarding anisotropy and extinction. This precomputation is independent of the data set and the transfer function, and results in a small preintegration table. During rendering, the look-up table is accessed for each ray sample point with respect to the viewing direction, phase function, and material properties in the spherical neighborhood of the sample. Our rendering technique is efficient and versatile because it readily fits in existing ray marching algorithms and can be combined with local illumination and volumetric ambient occlusion. It provides interactive volumetric scattering and soft shadows, with interactive control of the transfer function, anisotropy parameter of the phase function, lighting conditions, and viewpoint. A GPU implementation demonstrates the benefits of ambient scattering for the visualization of different types of data sets, with respect to spatial perception, high-quality illumination, translucency, and rendering speed.

The 3D visualization of astronomical nebulae is a challenging problem since only a single 2D projection is observable from our fixed vantage point on Earth. We attempt to generate plausible and realistic looking volumetric visualizations via a tomographic approach that exploits the spherical or axial symmetry prevalent in some relevant types of nebulae. Different types of symmetry can be implemented by using different randomized distributions of virtual cameras. Our approach is based on an iterative compressed sensing reconstruction algorithm that we extend with support for position-dependent volumetric regularization and linear equality constraints. We present a distributed multi-GPU implementation that is capable of reconstructing high-resolution datasets from arbitrary projections. Its robustness and scalability are demonstrated for astronomical imagery from the Hubble Space Telescope. The resulting volumetric data is visualized using direct volume rendering. Compared to previous approaches, our method preserves a much higher amount of detail and visual variety in the 3D visualization, especially for objects with only approximate symmetry.

Sort-first distributions have been studied and used far less than sort-last distributions for parallel volume rendering, especially when the data are too large to be replicated fully. We demonstrate that sort-first distributions are not only a viable method of performing data-scalable parallel volume rendering, but more importantly they allow for a range of rendering algorithms and techniques that are not efficient with sort-last distributions. Several of these algorithms are discussed and two of them are implemented in a parallel environment: a new improved variant of early ray termination to speed up rendering when volumetric occlusion occurs and a volumetric shadowing technique that produces more realistic and informative images based on half angle slicing. Improved methods of distributing the computation of the load balancing and loading portions of a subdivided data set are also presented. Our detailed test results for a typical GPU cluster with distributed memory show that our sort-first rendering algorithm outperforms sort-last rendering in many scenarios.

We present a parallel conjugate gradient solver for the Poisson problem optimized for multi-GPU platforms. Our approach includes a novel heuristic Poisson preconditioner well suited for massively-parallel SIMD processing. Furthermore, we address the problem of limited transfer rates over typical data channels such as the PCI-express bus relative to the bandwidth requirements of powerful GPUs. Specifically, naive communication schemes can severely reduce the achievable speedup in such communication-intense algorithms. For this reason, we employ overlapping memory transfers to establish a high level of concurrency and to improve scalability. We have implemented our model on a high-performance workstation with multiple hardware accelerators. We discuss the mathematical principles, give implementation details, and present the performance and the scalability of the system.

We extend direct volume rendering with a unified model for generalized isosurfaces, also called interval volumes, allowing a wider spectrum of visual classification. We generalize the concept of scale-invariant opacity—typical for isosurface rendering—to semi-transparent interval volumes. Scale-invariant rendering is independent of physical space dimensions and therefore directly facilitates the analysis of data characteristics. Our model represents sharp isosurfaces as limits of interval volumes and combines them with features of direct volume rendering. Our objective is accurate rendering, guaranteeing that all isosurfaces and interval volumes are visualized in a crack-free way with correct spatial ordering. We achieve simultaneous direct and interval volume rendering by extending preintegration and explicit peak finding with data-driven splitting of ray integration and hybrid computation in physical and data domains. Our algorithm is suitable for efficient parallel processing for interactive applications as demonstrated by our CUDA implementation.