Computer graphics processing and selective visual display system – Computer graphics processing – Three-dimension
Reexamination Certificate
2001-03-16
2003-09-30
Zimmerman, Mark (Department: 2671)
Computer graphics processing and selective visual display system
Computer graphics processing
Three-dimension
C345S424000, C345S590000
Reexamination Certificate
active
06628280
ABSTRACT:
FIELD OF THE INVENTION
The invention relates generally to sampled distance fields, and more particularly to adaptively resampling distance fields to obtain a more precise graphics model.
BACKGROUND OF THE INVENTION
Designing realistic digitized models is a major challenge for the animation industry, particularly when the models represent characters such as people—real or imagined, animals, and cartoon characters. Animation artists generally employ two production methods, alone or in combination. In one method, maquettes are sculpted from a traditional medium like clay and then digitized. In another method, the models are constructed using one of several commercial or custom computerized modeling systems, such as MAYA, SoftImage, 3DStudioMax, FormZ, and Houdini.
Clay is the medium of choice for most animation artists because it is expressive, and working with clay is intuitive. It is difficult to improve on clay as the ideal modeling medium. A standard approach for designing clay-based digital models involves sculpting a clay maquette and digitizing or scanning the maquette to generate the digital model. There is a plethora of systems for scanning objects and generating surface models from the scanned data.
However, sculpting with clay and then digitizing the clay maquette has several limitations for digital animation. Much detail can be lost in the digitizing process because scanners and digitizers have inherent limitations in resolution, are unable to digitize occluded or hard-to-reach surfaces, and are subject to noise. Thus, some of the advantages of clay are lost in the scanning process. Furthermore, long-term storage of the clay models is difficult. It is important to note, however, that scanned models can provide good first order approximations to the geometry of the clay maquettes that can then be enhanced by a computer-based modeling system. Hence, it is important that any modeling system does accept scanned data as input.
Most prior art computerized modeling systems typically use polygons, non-uniform rational B-splines (NURBS), or other subdivisions of planar surfaces to represent the shape of a model. However, all three representations have limitations.
Polygon models require a large number of vertices to represent detailed surfaces, particularly when the surfaces are highly curved, textured, or include sharp edges. This makes model generation and editing with polygons cumbersome and time consuming. Because NURBS are topologically restricted, they must be pieced together to form complex shapes. This presents numerous technical and interface challenges, see DeRose et al., “
Subdivision surfaces in character animation
,” Proc. SIGGRAPH '98, pp. 85-94, 1998. While subdivision of planar surfaces does not suffer from the same topological restrictions as NURBS, controlling shape changes, and adding fine detail during editing are difficult. As another problem, all of these modeling techniques are only surface representations—the models are nothing more than hollow shells and nothing is known about interior portions.
Even worse, sculpting of surface representations can lead to models that are not “watertight.” For example, seams where different resolution NURBS are joined can separate to form annoying cracks and holes. This means that the sculpted models have to be carefully inspected and edited before a watertight finished product is produced. Watertight models are important for several applications such as rapid prototyping technologies including stereo lithography.
All computerized modeling systems usually perform editing by manipulating control vertices. This requires significant skill and patience, as well as foresight and careful planning to ensure that the models have enough control vertices where detail is desired. For these reasons, computerized modeling systems do not rival the intuitive and expressive nature of clay.
To make the manipulation more intuitive, most modeling systems allow the user to interact with groups of control vertices using a computer implemented (digital) sculpting tool. For example, in MAYA Artisan, NURBS models are modified via a brush tool that manipulates groups of control vertices. Operations for pushing, pulling, smoothing, and erasing the surface are well known, and the brush tool can affect control vertices in a region of diminishing influence around its center, resulting in a softening of the sculpted shape. The amount of detail in the sculpted surface depends on the number of control vertices in the region of the sculpting tool. Finer control requires more subdivision of the NURBS surface, resulting in a less responsive system and a larger model. Often, mesh subdivision is user controlled and preset. It does not adapt to tool selection, or the detail of the sculpted surface. Hence, achieving fine detail in desired regions without excessive subdivision of the surface in other regions requires significant foresight and planning.
SoftImage provides a sculpting interface, called “Meta-Clay,” that is similar to the “metaballs” technology as described by Wyvill et al. in “
Animating soft objects
,” the Visual Computer, 2(4):235-242, 1986. Meta-Clay is a density based representation for modeling organic, sculpted objects. This representation produces blobby shapes, and does not represent edges, corners, or fine detail.
Sculpting of parametric models are described by Fowler, “
Geometric manipulation of tensor product surfaces
,” Proc. of the 1992 Symposium on Interactive 3D Graphics, pp. 101-108, 1992, Khodakovsky et al. “
Fine Level Feature Editing for Subdivision Surfaces
,” ACM Solid Modeling Symposium, 1999, and Terzopoulos et al. “
Dynamic NURBS with geometric constraints for interactive sculpting,
” ACM Transactions On Graphics, 13(2), pp. 103-136, 1994. However, each of these suffers from some of the limitations described above, and none attain sculpted results of the quality required for the animation industry.
To address the problems of polygon, NURBS, and other surface subdivision representations, volumetric data structures can be used. These data structures can be generated parametrically or by sampling techniques using, for example, laser or other ranging techniques, or scanners that penetrate the object to be modeled. Volumetric data can represent both the exterior and interior portions of models. The volumetric data are then sculpted by applying digital sculpting tools. The tools modify sample values near where the sculpting tool interacts with the volume. For these reasons, sampled volumes hold more promise as a data structure for digital clay.
FreeForm is a commercial system for sculpting volumetric models. FreeForm includes a three degree-of-freedom haptic input device which uses force feedback to provide the user with a sense of touch when sculpting. Models are represented as regularly sampled intensity values. This greatly limits the amount of detail that can be achieved, and requires excessive amounts of memory. For example, a minimum system requires 512 MB of random access memory (RAM). Intensity values can be low-pass filtered to reduce aliasing artifacts in the sculpted models, resulting in smoothed edges and rounded corners typical in volumetric sculpting systems.
To take advantage of standard hardware rendering engines, volumetric models can be converted to polygon models using a method such as described by Lorensen et al. in “
Marching Cubes: A High Resolution
3
D Surface Construction Algorithm
,” Proc. SIGGRAPH '87, pp.163-169, 1987. However, with large volume sizes, Marching Cubes produces an excessive number of triangles, leading to memory overload and bottlenecks in the graphics rendering pipeline which limit interactivity.
There are a number of publications that describe volume-based sculpting systems including Avila et al. “
A haptic interaction method for volume visualization,
” Proc. IEEE Visualization'96, pp. 197-204, 1996, Baerentzen, “
Octree
-
based volume sculpting Proc. Late Breaking Hot Topics,
” IEEE Visualization'98, pp. 9-12, 1998, Galyean et al. “
Sculpting: An Interactive Vo
Frisken Sarah F.
Perry Ronald N.
Brinkman Dirk
Mitsubishi Electric Research Laboratories Inc.
LandOfFree
Method for selectively regenerating an adaptively sampled... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for selectively regenerating an adaptively sampled..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for selectively regenerating an adaptively sampled... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3037187