Disney Enterprises, Inc.

Pixar

Peter-Pike Sloan

Wojciech Jarosz

Derek Nowrouzezahrai

Robert Thomas

Matthias Zwicker

Robert L Cook

Manuel Kraemer

Lakulish Shailesh Antani

Ladislav Kavan

Pitchaya Sitthi-Amorn

Hong Yuan

A method for rendering radiance for a volumetric medium is provided. A photon simulation produces a representation of photon beams in a scene. The photon beams are rendered with respect to a camera viewpoint, by computing an estimated radiance associated with the photon beams. A global radius scaling factor can be applied to obtain different radii for the photon beams. Over multiple applications of these steps, the global radius scaling factor can be decreased, thereby reducing overall error by facilitating convergence. Finally, the renderer can be efficiently implemented on the GPU as a splatting operation, for use in interactive and real-time applications.

Image processing using progressive generation of intermediate images using photon beams of varying parameters

Surfaces without a global surface coordinate system are divided into surface regions having local surface coordinate systems to enable the caching of surface attribute values. A surface attribute value for a surface region may include contributions from two or more adjacent surfaces. Sample points may be arranged at the corners, rather than centers, of surface regions and include prefiltered values based on two or more surfaces. A renderer may sample the surface attribute function using these prefiltered values without accessing any adjacent surfaces, even if the renderer's filter crosses a surface boundary. A multiresolution cache stores surface attribute values at different resolution levels for surface regions of one or more surfaces, which may be discontiguous. Two or more resolution levels may have the same number of sample points but have values based on filters with different areas and spatial frequency limits. Resolution levels may be selected based on geodesic distance on a surface.

Caching attributes of surfaces without global parameterizations

A technique for physically-based cloth simulation uses linear upsampling operators. The upsampling operators enrich the appearance of a coarse mesh physical cloth simulation. The technique starts by pre-computing the upsampling operators using a pair of coarse and fine training simulations aligned with tracking constraints using harmonic test functions. Then the upsampling operators are trained using a novel regularization technique that enables mid-scale detail learning without over-fitting. Oscillatory modes may be introduced to add dynamic details not captured by the coarse mesh simulation alone. Trained upsampling operators can then be advantageously applied to coarse mesh simulations of cloth to add realistic detail to the cloth in real-time three-dimensional applications.

Physics-inspired up sampling for cloth simulation

Techniques are disclosed for using a local lighting representation to explicitly model spatial variation of a character in a graphics scene as well as for using error driven criteria to determine whether to evaluate a given light source analytically or in a lighting rig. For near light sources, the error driven criteria may be used to determine when a spherical light source should be evaluated in the lighting rig verses analytically. For large characters, local irradiance models may be used to provide a limited form of spatial variation.

Irradiance rigs

Embodiments of the invention provide techniques for calculating volumetric obscurance. Volumetric obscurance is an alternative approach to screen-space ambient occlusion that can be efficiently evaluated on graphics processing units (GPUs). In one embodiment, to calculate the volumetric obscurance at a particular pixel location, an integral is calculated that sums distances between depth values in a depth buffer relative to a surface of a sphere centered at the pixel location. In alternative embodiments, a statistical model of the depth buffer can be used to calculate the volumetric obscurance.

System and method for modeling ambient occlusion by calculating volumetric obscurance

A method and system for progressively rendering radiance for a volumetric medium is provided. A photon simulation produces a representation of photon beams in a scene. The photon beams are rendered with respect to a camera viewpoint, by computing an estimated radiance associated with the photon beams. A global radius scaling factor is applied to the photon beams. Over multiple iterations of these steps, the global radius scaling factor is progressively decreased, thereby reducing overall error by facilitating convergence. This order can be mixed up. Finally, the renderer can be efficiently implemented on the GPU as a splatting operation, for use in interactive and real-time applications.

Image processing using iterative generation of intermediate images using photon beams of varying parameters

Modular radiance transfer pre-computes a spectral decomposition of a one bounce transport operator for each of a set of shapes and replaces a scene dependent direct-to-indirect transfer lighting computation with an art-directable mapping step. Modular radiance transfer provides an approach for rendering an approximate indirect lighting which requires small amounts of data, can be used to model very large (or even random scenes), and propagates much of the indirect lighting using a small number of spectral coefficients, so that rendering scales well to large scenes.

Modular radiance transfer

Techniques are disclosed to determine the irradiance from a disk light source with much higher accuracy than may be obtained using the common approximation often used currently. By pre-computing, SH coefficients using a specific coordinate frame, only six SH coefficients need to be stored in a first and second texture, one used to determine irradiance of a point within the radius of a disk light source and another for more distant points.

Growing community of inventors

Cottonwood Heights, UT, United States of America

Peter-Pike Sloan