scispace - formally typeset
Search or ask a question

Showing papers in "Journal of Graphics Tools in 2004"


Journal ArticleDOI
TL;DR: A new algorithm for digital inpainting based on the fast marching method for level set applications is presented, very simple to implement, fast, and produces nearly identical results to more complex, and usually slower, known methods.
Abstract: Digital inpainting provides a means for reconstruction of small damaged portions of an image. Although the inpainting basics are straightforward, most inpainting techniques published in the literature are complex to understand and implement. We present here a new algorithm for digital inpainting based on the fast marching method for level set applications. Our algorithm is very simple to implement, fast, and produces nearly identical results to more complex, and usually slower, known methods. Source code is available online.

1,109 citations


Journal ArticleDOI
TL;DR: This work considers intersections between rays and the simplest parametric surface, the bilinear patch, and presents a complete, efficient, robust, and graceful formulation to solve ray-bilinearpatch intersections quickly.
Abstract: Ray tracing and other techniques employ algorithms which require the intersection between a three-dimensional parametric ray and an object to be computed. The object to intersect is typically a sphere, triangle, or polygon but many surface types are possible. In this work we consider intersections between rays and the simplest parametric surface, the bilinear patch. Unlike other surfaces, solving the ray-bilinear patch intersection with simple algebraic manipulations fails. We present a complete, efficient, robust, and graceful formulation to solve ray-bilinear patch intersections quickly. Source code is available online.

48 citations


Journal ArticleDOI
TL;DR: A cloud system that extends texture splatting on particles to model a dozen cloud types, an improvement over earlier systems that modeled only one type of cumulus, and introduces a way to simulate cloud form ation and dissipation using texture splatted particles.
Abstract: Clouds are an important aspect of rendering outdoor scenes. This paper describes a cloud system that extends texture splatting on particles to model a dozen cloud types (e.g., strat us, cumulus congest us, cumulonimbus), an improvement over earlier systems that modeled only one type of cumulus. We also achieve fast realtime rendering, even for scenes of dense overcast coverage, which was a limitation for previous systems. We present a new shading model that uses artist-driven controls rather than a programmatic approach to approximate lighting. This is suitable when fine-grained control over the look-and-feel is necessary, and artistic resources are available. We also introduce a way to simulate cloud form ation and dissipation using texture splatted particles.

44 citations


Journal ArticleDOI
TL;DR: This method tests the ray against the edges comprising the silhouette of the box instead of testing against individual faces, which simplifies the technique and shows the technique's performance is up to 93% faster than traditional methods if an intersection distance is not needed.
Abstract: Fast ray-axis aligned bounding box overlap tests can be performed by utilizing Pliicker coordinates. This method tests the ray against the edges comprising the silhouette of the box instead of testing against individual faces. Projection of the edges onto a two-dimensional plane to generate the silhouette is not necessary, which simplifies the technique. The method is division-free and successive calculations are independent and consist simply of dot product operations, which permits vectorization. The method does not compute an intersection distance along the ray to the box, but this can be added as an additional step. Storage of Pliicker coordinates is unnecessary, permitting integration into existing systems. Test results show the technique's performance is up to 93% faster than traditional methods if an intersection distance is not needed.

42 citations


Journal ArticleDOI
TL;DR: The implementation of a photogrammetric reconstruction tool used on the post-production of Matrix Reloaded and Matrix Revolutions that was built using automatic differentiation is described.
Abstract: Differential calculus is ubiquitous in digital movie production. We give a novel presentation of automatic differentiation, a method for computing derivatives of functions, that is not well known within the graphics community, and describe some applications of this method. In particular we describe the implementation of a photogrammetric reconstruction tool used on the post-production of Matrix Reloaded and Matrix Revolutions that was built using automatic differentiation.

39 citations


Journal ArticleDOI
TL;DR: A perceptually based image comparison processtliat can be used to tell when images are perceptually identical even though they contain imperceptible numerical differences.
Abstract: This paper describes a perceptually based image comparison processtliat can be used to tell when images are perceptually identical even though they contain imperceptible numerical differences. The technique has shown much utility in the production testing of rendering software.

29 citations


Journal ArticleDOI
TL;DR: It is shown that the relationship between s and the number of triangles in a model, f, is bounded above by s = O(f) and below by s= Ω(1), and that the expected value of s over all observation points at infinity is proportional to the sum of the dihedral angles.
Abstract: Silhouettes have many applications in computer gr aphics such as non-photorealistic edge rendering, fur rendering, shadow volume creation, a nd anti-aliasing. The number of edges, s, in the silhouette of a model observed from a point is ther efore useful in analyzing such algorithms. This paper examines, from a theoretical viewpoi nt, a menagerie of objects with interesting silhouettes (including those with minimal and maximal silhouettes). It shows that the relationship between and s and the number of triangles in a model, f, is bounded above by s = O( f) and below by s = Ω(1), and that the expected value of s over all observation points at infinity is proport ional to the sum of the dihedral angles. In practice, the models used with silhouette-ba sed rendering algorithms are triangle meshes that a re manually constructed or result from scans of human- made objects. They consist of only surface geometry with few cracks; there is no internal deta il like the engine under a car's hood. Geometric and aesthetic constraints on these models appear to create an inherent relationship between f and s. Measurements of the actual silhouettes of real-worl d 3D models with polygon counts varied across six orders of magnitude show them to follow the relatio nship s ~ f 0.8 . Furthermore, the expected value of s at infinity is a good approximation of the expecte d silhouette size for a viewer at a finite location.

26 citations


Journal ArticleDOI
TL;DR: A dynamic adaptive mesh library which maintains a conforming triangulation of time-varying surfaces, which facilitates the development of graphics applications that manipulate triangulated surfaces.
Abstract: In this paper, we present a dynamic adaptive mesh library which maintains a conforming triangulation of time-varying surfaces. The user supplies an initial mesh, a surface sampling procedure, and a set of adaptation criteria. The mesh is automatically modified in order to conform to user-defined characteristics, while the surface changes over time. The mesh representation is based on a half-edge data structure without any extra storage requirements. The mesh has an underlying semiregular multiresolution structure. Furthermore, the specification of desired mesh characteristics can be based on very general adaptation rules. This scheme facilitates the development of graphics applications that manipulate triangulated surfaces. The library source code is available online.

21 citations


Journal ArticleDOI
TL;DR: Algorithms for computing the minimal distance between a point and a cylinder, a cone, a line-swept sphere, and a cone-sphere are presented.
Abstract: This paper presents algorithms for computing the minimal distance between a point and a cylinder, a cone, a line-swept sphere, and a cone-sphere. Some optimizations are provided when queries are performed along a line which may be useful for voxelization applications. Source code is available online.

21 citations


Journal ArticleDOI
TL;DR: A new, fast approach for updating oriented bounding box hierarchies for articulated humanoid models, using a bottom-up approach that approximates existing techniques by assuming a major body axis.
Abstract: We present a new, fast approach for updating oriented bounding box hierarchies for articulated humanoid models, using a bottom-up approach. The algorithm approximates existing techniques by assuming a major body axis. Existence of a major axis allows merging of bounding boxes in a hierarchy approximately but with sufficient accuracy. For scenarios with close proximity, a speed-up by a factor 2 on average is achieved compared to existing techniques.

16 citations


Journal ArticleDOI
TL;DR: This paper examines the implementation of virtual goniphotometers used to obtain BRDF and BTDF records from algorithmic scattering models and discusses practical issues, namely the subdivision of the devices' collector sphere and the ray density required to obtain reliable BRD F and BT DF estimates.
Abstract: Many scattering models have been presented in the graphics literature Few of them, however, have been evaluated through comparisons with real measured data As the demand for plausible and predictable scattering models increases, more attention is given to performing such comparisons In this paper, we examine the implementation of virtual goniphotometers used to obtain BRDF (Bidirectional Reflect ance Distribution Function) and BTDF (Bidirectional Transmittance Distribution Function) records from algorithmic scattering models These records can be compared to data from actual experiments in order to validate the models Our discussion focuses on practical issues, namely the subdivision of the devices' collector sphere and the ray density required to obtain reliable BRD F and BTDF estimates The subdivision techniques examined in this paper have been used before in publications, but the details of their computation are not readily available in the literature Although the mathematical bound pres

Journal ArticleDOI
TL;DR: A new subdivision schème that can be used to generate limit surfaces of either odd or even, degree, differs from existing evendegree schemes in that it is able to correctly propagate attributes such as texturecoordinates or polygon colors to intermediate meshes without introducing errors.
Abstract: We introduce a new subdivision scheme that can be used to generate limit surfaces of either odd or even, degree. The scheme, which requires only linear subdivision and local smoothing operations, differs from existing evendegree schemes in that it is able to correctly propagate attributes such as texturecoordinates or polygon colors to intermediate meshes without introducing errors. The new approach also permits the use of creases on even-degree surfaces, in a similar fashion to current odd-degree methods. The new scheme also permits the generation of surfaces that interpolate between limit-surfaces of different degrees. In addition, the target limit-surface degree can be specified to vary over the input mesh; the algorithm smoothly interpolates between regions of differing degree. As a result, artists have more flexible tools for controlling the limit-surface shape.

Journal ArticleDOI
TL;DR: This paper explores the methods used to warp the skin around the transition from the shape-shifting character in X-Men 2, Mystique, using Photorealistic Renderman, by extrapolating a two-channel vector map from the single-channel texture map that was used to drive Mystique's transition.
Abstract: Abstract This paper explores the methods used to warp the skin around the transition from the shape-shifting character in X-Men 2, Mystique, using Photorealistic Renderman. We begin by extrapolating a two-channel vector map from the single-channel texture map that was used to drive Mystique's transition. We show how to use the vector maps to warp the texture across borders in separate UV spaces. Finally, we show how to solve the problem of antialiasing across the separate UV spaces.

Journal ArticleDOI
TL;DR: The Journal of Graphics Tools: Vol. 9, Special Issue on Computer Graphics for Motion Picture Production, pp. 1-1 as discussed by the authors, is a special issue on computer graphics for motion picture production.
Abstract: (2004). Special Issue on Computer Graphics for Motion Picture Production. Journal of Graphics Tools: Vol. 9, Special Issue on Computer Graphics for Motion Picture Production, pp. 1-1.

Journal ArticleDOI
TL;DR: The proposed stratified sampling technique for an ellipse is stratified, efficient, and feasible for the computer graphics community and has proven useful in realistic image synthesis using Monte Carlo direct lighting and Monte Carlo path tracing.
Abstract: We present an efficient stratified sampling technique for an ellipse. Our technique provides extensions of pre vious work for sampling a disk with lowdistortion; it works for both polar and concentric maps between a square and an ellipse. The mapping preserves adjacency and fractional area, and does not require any numerical computation. As in the previous work, the concentric map demonstrates better performance. Our technique has proven useful in realistic image synthesis using Monte Carlo direct lighting and Monte Carlo path tracing. In conclusion, the proposed technique is stratified, efficient, and feasible for the computer graphics community. Source code is available online.

Journal ArticleDOI
TL;DR: Maya's built-in file-referencing mechanism provides a small step towards enabling finer grain data management, but falls short in enabling either of the potential key advantages outlined above.
Abstract: Systems for effective management of shot content data are critical to the support of any large-scale digital production. In regards to the data's structure, there are a variety of advantages in maintaining and managing data in a granular form, as a collection of separable data assets. Two key abilities afforded through data management in a granular form are 1) an ability to selectively load a subset of shot assets for improved performance with a reduced memory footprint, and 2) an ability to selectively lock a subset of shot assets enabling parallel shot work by multiple artists. Both translate directly into improved artist and production pipeline efficiencies. Maya's built-in file-referencing mechanism provides a small step towards enabling finer grain data management, but falls short in enabling either of the potential key advantages outlined above. The Maya Assets system was developed at DreamWorks Animation, Inc. in support of these capabilities and other such efficiency improvements. The sys...