Abstract: A method comprising generating a surface detail model using a modeling technique, and rendering surface detail in accordance with the developed surface detail model over an object surface. The described method enables computer-generated images containing representations of hair to be rendered in real time.

Abstract: A method comprising generating a surface detail model using a modeling technique, and rendering surface detail in accordance with the developed surface detail model over an object surface. The described method enables computer-generated images containing representations of hair to be rendered in real time.

Abstract: A method comprising generating a surface detail model using a modeling technique, and rendering surface detail in accordance with the developed surface detail model over an object surface. The described method enables computer-generated images containing representations of hair to be rendered in real time.

Abstract: A space-time level of detail coder converts a time dependent geometry stream into a hierarchical structure, including levels of detail in the space and time dimensions, and expansion records. The expansion records specify how to reconstruct a mesh from deltas representing differences between levels of detail. For low bandwidth transmission, only selected expansion records need be sent depending on the appropriate level of detail of the object being represented for the current time sample.

Abstract: Methods for coding a time-dependent geometry stream include a basis decomposition coder and a column/row prediction coder. The basis decomposition coder uses principal component analysis to decompose a time dependent geometry matrix into basis vectors and weights. The weights and basis vectors are coded separately. Optionally, the residual between a mesh constructed from the weights and basis vectors and the original mesh can be encoded as well. The column/row predictor exploits coherence in a matrix of time dependent geometry by encoding differences among neighboring rows and columns. Row and column sorting optimizes this form of coding by re-arranging rows and columns to improve similarity among neighboring rows/columns.

Abstract: Methods and arrangements are provided for real-time rendering of scenes having various light sources and objects having differing specular surfaces. An offline encoder is employed to parameterize images by two or more arbitrary variables allowing view, lighting, and object changes. The parameterized images are encoded as a set of per-object parameterized textures based on shading models, camera parameters, and the scene's geometry. Texture maps are inferred from a ray-tracer's segmented imagery to provide the best match when applied to specific graphics hardware. The parameterized textures are encoded as a multidimensional Laplacian pyramid on fixed size blocks of parameter space. This technique captures the coherence in parameterized animations and decodes directly into texture maps that are easy to load into conventional graphics hardware.

Abstract: A geometric transform coder encodes the residual between a transformed base mesh and the original mesh, for each of a series of meshes in a time dependent geometry stream. The geometric transform that matches the base mesh to the current mesh can be derived from the input stream or provided by the author of the model.

Abstract: A method comprising generating a surface detail model using a modeling technique, and rendering surface detail in accordance with the developed surface detail model over an object surface. The described method enables computer-generated images containing representations of hair to be rendered in real time.

Abstract: A decision-theoretic regulator employs a method for allocating computational resources to components of media content to create the highest quality output for a budget of rendering resources. The components of the content represent parts of the content that have independent quality parameters that the regulator can vary to trade-off quality for computational savings. For example, in multimedia content, the components might be objects in a 3D graphics scene. The method allocates computational resources by attempting to minimize the total expected cost of a rendering task. The method computes the raw error for a rendering action on a component and then maps the raw error to a perceived error based on empirical evidence of how users perceive errors in rendered output. The expected cost is computed from the perceived error or raw error by applying a model of attention that gives the probability that a user is focusing his or her attention on a component.

Abstract: An image generator takes graphical objects and an occlusion relationship for the objects and resolves non-binary occlusion cycles with image compositing operations to produce an output image of the objects. The image generator takes an occlusion relationship for objects in a scene and a set of antialiased image layers with transparency of the objects and produces an antialiased image of the objects with hidden surfaces eliminated. One implementation operates on subsets of the objects in a scene that form non-binary cycles. This implementation uses a chain of atop operators to combine occluding objects with a selected object from a subset, and then combines this result with other objects in the cycle using over image operations. Another implementation computes a chain of out image operations for each object to combine the image layers of the occluding objects with the image layer of the object. The results of each chain of out image operations are summed to produce an output image.

Abstract: A layered graphics rendering pipeline measures image fidelity ("fiducials") to determine how accurately a transformed image layer approximates a rendering of a 3D object. The graphics rendering pipeline approximates the change in position or color of 3D object by transforming a rendering of the 3D object from a previous frame. The pipeline uses the fiducials to control rendering of factored scene elements to independent image layers. The pipeline then combines the layers to compute frames of animation. The types of fiducials include sampling, visibility, and photometric fiducials. The sampling fiducial measures the distortion of an image sample when warped to screen coordinates. The visibility fiducial measures the change in visibility of a scene element since a previous rendering of the scene element.

Abstract: A layered graphics rendering pipeline for real time 3D animation independently renders terms in a shading model to separate image layers. The layered pipeline factors the shading model into separate image layers and renders geometry to these layers independently. Each layer can have an independent update rate and a spatial resolution different than the resolution of the output images. A compositor that supports one or more image operators composites the factored layers into an output image to generate frames of animation. To reduce rendering overhead, factored terms can be rendered once and then re-used in later frames by warping the initial rendering.

Abstract: In an image processing system, a method for generating a images includes rendering graphical models comprising a scene to separate image layers called "gsprites," and then compositing these image layers to generate an image. An image processor can retrieve gsprites from memory, transform them, and composite them for display at video rates. Gsprites can be re-rendered or updated at different rates. Reducing the rendering overhead of the system, the image processor can perform an affine transformation on the gsprite to approximate motion of the graphical object that it represents, rather than re-render the object. Objects in a scene can be queued for re-rendering based on a predefined update rate, or based on the accuracy of representing the object with a transformed gsprite, rendered for a previously displayed image.