Naylor BF. Partitioning tree image representation and generation from 3D geometric models. Proceedings of Graphics Interface 1992. p. 201}12.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....since instead of comparing the scene with each of the N shadow frusta, it is compared with one tree with an O(logN) expected depth (potentially O(N) while taking in into account occluder fusion. The above technique is conservative; an alternative exact method was proposed much earlier by Naylor [67]. That involved a merging of the occlusion tree with the BSP tree representing the scene geometry. 6 Point based image precision techniques As the name suggests image precision algorithms perform the culling at the discrete representation of the image. The key feature in these algorithms is that ....

Bruce F. Naylor. Partitioning tree image representation and generation from 3D geometric models. In Proceedings of Graphics Interface '92, pages 201--212, 1992.

....Besides approaches which use a building floor plan for this purpose [1, 2, 26, 18] no other methods for deriving decomposition hierarchies from CAD models are known to us. A more general approach is to organize a polygonal model into regular spatial decomposition schemes, such as BSP trees [8, 20, 11] or Octrees [10, 12, 6, 7, 25] While these decomposition schemes produce good results on polygonal models extracted by the Marching Cubes algorithm from uniform grid volume datasets which provide a natural decomposition on Marching Cubes cell base , these schemes run into numerous ....

B. Naylor. Partitioning Tree Image Representation and Generation From 3D Geometric Models. In Proc. of Graphics Interface, pages 201--212, 1992.

....for such scenes are view frustum culling, visibility culling and level of detail switching [13] However, little work has been done so far concerning the generalization of these methods for dynamic models, which are the bread and butter of VR applications. Consider visibility culling techniques [6, 10, 11, 15], also known as occlusion culling or output sensitive visibility algorithms. These techniques utilize the occlusions in a scene to render it in time proportional to the number of visible objects, rather than the total number of objects, which can be greater by orders of magnitude. However, these ....

....culling for static scenes A few visibility culling algorithms have been presented in recent years. All these algorithms construct a spatial hierarchical data structure as a preprocessing stage: a binary space partitioning tree (BSP tree) in Naylor s partitioning tree visible surface algorithm [15]; an octree or a k D tree in Greene et al. s hierarchical Z bu#er algorithm [11] and in Coorg and Teller s object space visibility algorithm [6] Given a viewpoint, these algorithms proceed by traversing the spatial hierarchical data structure in a top down, near to far order, terminating the ....

[Article contains additional citation context not shown here]

B. F. Naylor. Partitioning tree image representation and generation from 3D geometric models. In Proceedings of Graphics Interface '92, pages 201-- 212, Vancouver, May 1992.

....of other objects. The BSP tree identifying an object has all the constituent polygon fragments at the leaf nodes. Binary Space Partition tree representing the complete scene is constructed by successive merging of objects. Merging operation is developed on the work done by Thibault and Naylor[2,4] for representation of arbitrary polyhedral solids and set operations on polyhedra. We give simpler and ecient dynamic updates, especially that of deletion, by using a separating plane between two objects. The separating 1 plane methodology has been described in [9] This version of the BSP tree ....

Naylor, B., Partitioning tree image representation and generation from 3D geometric models, Proceedings of Graphics Interface '92, 201-212, may 1992.

....method is therefore suitable for walk through applications. However, it is not suitable for interactive modification of objects in the scene, since any change in an object s position could destroy the ordering and may require the reconstruction of the shadow tree. Similar structures were used in [10] for image representation and in [4, 2] for determining illumination discontinuities from area light sources. The algorithm presented here employs a generalization of the SVBSP tree that does not require the construction of a BSP tree of the original scene polygons, and that does support near ....

B. F. Naylor. Partitioning tree image representation and generation from 3d geometric models. In Proceedings of the Graphics Interface '92, pages 201-- 212, 1992.

....assistance in design of architectural structures and solid modeling. We solve the dynamic hidden surface removal problem using a variation of the traditional Binary Space Partition (BSP) Trees. The BSP tree is a versatile structure commonly used for hidden surface removal and solid modeling [4,5,8]. Other uses of the BSP Trees are in shadow generation, collision detection, implementing set operations on objects etc [3,4,5,6] A BSP tree is a binary tree used to partition n dimensional space and essentially to sort polytopes. The tree taken as a whole represents the entire space, and each ....

....using a variation of the traditional Binary Space Partition (BSP) Trees. The BSP tree is a versatile structure commonly used for hidden surface removal and solid modeling [4,5,8] Other uses of the BSP Trees are in shadow generation, collision detection, implementing set operations on objects etc [3,4,5,6]. A BSP tree is a binary tree used to partition n dimensional space and essentially to sort polytopes. The tree taken as a whole represents the entire space, and each subtree represents a convex subspace. Each node stores a hyperplane which divides the space it represents into two parts, and has ....

Naylor, B., Partitioning tree image representation and generation from 3D geometric models, Proceedings of Graphics Interface '92, 201--212, may 1992.

....occlusion culling overhead. Therefore these methods are often used e.g. in games [1] where the frame rate is the major criterion and the time expensive precomputation does not hurt so much. Methods that, like our new method, do their occlusion calculations on the fly during the display phase [21,23,24,34] have the advantage that they do not need a time expensive precomputation but of course the occlusion calculation during the display phase produces some overhead. The hierarchical z buffer [13,31] is an image based on the fly method that uses a pyramid of z values to cull objects in large already ....

B. Naylor. Partitioning tree image representation and generation from 3d geometric models. Graphics Interface 92 pp. 201-212

....scenes. The BSP based visibility algorithm has been improved by Gordon and Chen to achieve output sensitivity of the algorithm for certain type of environment [74] An interesting output sensitive hidden surface removal algorithm for scenes described solely as a BSP tree was introduced by Naylor [126]. BSP trees have also been used to represent a collection arbitrary polyhedra and to perform set operations on the polyhedra [194, 122, 124, 125, 123] Dynamic changes to BSP trees were studied in [197, 38] Recently, Naylor introduced an algorithm for converting discrete images to 2D BSP trees ....

B. F. Naylor. Partitioning tree image representation and generation from 3D geometric models. In Proceedings of Graphics Interface '92, pages 201--212, May 1992.

....a technique by Meagher [20] that stores the scene in an octree, and the framebuffer in a quadtree. Meagher renders the octree in a strict front to back order, while keeping track of which parts of the quadtree get filled, in order to avoid touching parts of the octree that can not be seen. Naylor [22] proposes another version of this idea, where instead of using an octree and a quadtree, he uses two binary space partitioning trees [12] one in 3D, the other in 2D, to efficiently keep both the scene and the image respectively. The 3D BSP can be used to traverse the scene in a strict ....

B. F. Naylor. Partitioning tree image representation and generation from 3D geometric models. In Proceedings of Graphics Interface '92, pages 201--212, May, 1992. 12 Efficient Conservative Visibility Culling Using The Prioritized-Layered Projection Algorithm 13

....tree recursively. This method has an advantage over [39] in that instead of comparing the scene with each of the N shadow frusta, it is compared with one tree of depth (potentially) O(N) The above technique is conservative; an alternative exact method was proposed much earlier by Naylor [53]. That involved a merging of the occlusion tree with the BSP tree representing the scene geometry. 6.4 Prioritized layered projection Prioritized Layered Projection (PLP) is a technique for fast rendering of high depth complexity scenes. It works by estimating the visible polygons of a scene ....

....some sort of occlusion preserving simplification algorithm) to be able to generate the occlusion maps. Another interesting issue is how to deal with dynamic scenes. The more preprocessing used, the more expensive it is to deal with dynamic environments. The BSP tree method introduced by Naylor [53] already in 1992 can be thought of as somewhere between image precision and object precision, since although he used a 2D BSP tree in image space for culling the 3D scene, this was done using object precision operations rather than image precision. 8 From region visibility In a typical ....

B. F. Naylor. Partitioning tree image representation and generation from 3D geometric models. In Proceedings of Graphics Interface '92, pages 201--212, 1992.

....scenes, it is likely that most invisible triangles are occluded by a small subset of the triangles. Since computing this subset is exactly the same as the hidden surface removal problem itself, the authors present heuristics to estimate good occluders. A related algorithm of Naylor [77] processes the triangles in front to back order and uses a BSP in the image plane to represent the projected shadows. The disadvantage with all these algorithms is that they do not handle the case when a cell is invisible because it is occluded by multiple, disconnected triangles. See Figure ....

....the front toback order of the triangles. When the front to back order is not easily determined, he outlines a lazy z buffering approach to resolve visibility. Zhang et al. simply check the visibility of each moving object in each frame. Sudarsky and Gotsman [97, 98] extend Naylor s technique [77] to handle moving objects. They use the interesting idea of enclosing each moving object in a temporal bounding volume based on known estimates on the objects s velocity. The store the temporal bounding volume for each moving object in a BSP. Their algorithm attempts to process a moving object ....

B. F. Naylor, Partitioning tree image representation and generation from 3D geometric models, Proc. Graphics Interface '92, 1992, pp. 201--212.

....techniques require a preprocessing step and are limited to static environments which is the case with most potentially visible set methods (see below) Also many techniques are only meant to work with scenes that are densely occluded or that have some special geometric characteristics. Naylor [13] presented a method which realizes occlusion culling by generating a 2D BSP tree of the image from a 3D BSP tree of the scene. Occlusion culling methods that are specialized for 2D scenes use the simplicity of the geometry of such scenes which is based on a ground plan [14] 19] Potentially ....

B. Naylor. Partitioning tree image representation and generation from 3d geometric models. Graphics Interface 92 pp. 201-212, 1992

.... some techniques to overcome these drawbacks are discussed by Greene [19] Naylor presented an algorithm, based on a 3D BSP tree for the representation of the scene, a 2D BSP tree as image representation, and an algorithm to project the 3D BSP tree subdivided scene into the 2D BSP tree image [31]. Hong et al. proposed a fusion between the hierarchical z buffer algorithm [21] and the PVS algorithm in [27] In this z bufferassisted occlusion culling algorithm, a human colon is first subdivided into a tube of cells in a pre process. Thereafter, the occlusion is determined on the fly by ....

....In [34] Snyder and Lengyel proposed that the designer of the scene needs to provide the subdivision. Similarly, Zhang et al. used a pre defined scene database [40] The most general approach is to subdivide a polygonal model into more or less regular spatial subdivision schemes, such as BSP trees [14, 31, 19], k D trees [5] or Octrees [18, 21] While these subdivision schemes produce good results on polygonal models extracted by the Marching Cubes algorithm [26] from uniform grid volume datasets which provide a natural subdivision on Marching Cubes cell base, these schemes run into numerous ....

B. Naylor. Partitioning Tree Image Representation and Generation From 3D Geometric Models. In Proc. of Graphics Interface, pages 201--212, 1992.

....octree of BSP trees limits the application of this algorithm to static scenes. Naylor presented an algorithm, based on a 3D BSP tree for the representation of the scene, a 2D BSP tree as image representation, and an algorithm to project the 3D BSP tree subdivided scene into the 2D BSP tree image [20]. This approach can be considered as a generalization of Greene s hierarchical z buffer algorithm [14] Hong et al. proposed a fusion between the hierarchical z buffer algorithm [14] and the PVS algorithm in [18] In this z bufferassisted visibility algorithm, a human colon is first subdivided ....

B. Naylor. Partitioning tree image representation and generation from 3d geometric models. In Proc. of Graphics Interface '92, pages 201--212, 1992.

....of other objects. The BSP tree identifying an object has all the constituent polygon fragments at the leaf nodes. Binary Space Partition tree representing the complete scene is constructed by successive merging of objects. Merging operation is developed on the work done by Thibault and Naylor[2,4] for representation of arbitrary polyhedral solids and set operations on polyhedra. We give simpler and efficient dynamic updates, especially that of deletion, by using a separating plane between two objects. The separating plane methodology has been described in [9] This version of the BSP tree ....

Naylor, B., Partitioning tree image representation and generation from 3D geometric models, Proceedings of Graphics Interface '92, 201--212, may 1992.

....viewpoint are 10 traced via a depth first traveral through the winged pair structure, while corresponding convex regions of the visibility map are partitioned recursively. The algorithm is based on recursive beam tracing methods [23, 16, 35] and it is related to recursive convex decompositions [30]. The key feature is that topological information stored in the winged pair data structure (edge face adjacencies) is used to construct a visibility map with topological information and explicit silhouette edges. Psuedocode for the beam tracing algorithm is shown in Figure 6. During the traversal ....

Bruce F. Naylor. Partitioning tree image representation and generation from 3D geometric models. In Proceedings of Graphics Interface '92, pages 201--212, May 1992. 20

....by a limited portion of the model and a limited number of dynamic object motions. Therefore, outputsensitivity can reduce the amount of information that needs to be communicated from server to client. Output sensitive rendering is achieved by two complementary techniques: occlusion culling [8] [9], 10] 11] 12] 13] and level of detail control [14] 15] 16] Occlusion culling (also called visibility culling or output sensitive visibility calculation) finds the visible parts of the scene without wasting time on the occluded parts, not even to determine that they are occluded. ....

....for example, by physical simulation or by a user interface. Given these constraints, occlusion culling algorithms can be adapted to scenes containing multiple dynamic objects. This adaptation is shown for two algorithms: Greene et al. s hierarchical Z buffer [10] and Naylor s BSP tree projection [9]. Experimental results are given for both. 2 RELATED WORK 2.1 Static Scene Occlusion Culling The earliest known mention of occlusion culling was by Clark [19] He proposed using a hierarchical object representation, with a bounding sphere and a simply shaped bounded volume at each node of the ....

[Article contains additional citation context not shown here]

 B.F. Naylor, "Partitioning Tree Image Representation and Generation from 3D Geometric Models," Proc. Graphics Interface `92, pp. 201--212, Vancouver, May 1992.

....Bell Laboratories. y Bell Laboratories, 700 Mountain Avenue, 2A 202, Murray Hill, NJ 07974. Email: funk bell labs.com model be consistently and correctly oriented. More sophisticated rendering algorithms perform visibility culling by processing the interiors of the solid objects in the model [19, 20, 27]. In the case of interactive collision detection, some algorithms first process free space, i.e. the complement of the union of all the obstacles in the environment [11] Such algorithms require a correct representation of the boundary of the obstacles so that they can effectively construct the ....

Bruce F. Naylor. Partitioning tree image representation and generation from 3D geometric models. In Proc. Graphics Interface '92, pages 201--212, 1992.

....sorting of primitives. Once the primitives are ordered in depth, they can be rendered back to front for correct visibility. The binary space partitioning (BSP) algorithms [FKN80] produce an organization of polygons from which their depth ordering can be quickly derived. More recently, Naylor ( Nay92] described a algorithm which projects a 3 D BSP tree in object space into a 2D BSP tree in screen space for depth ordering and occlusion culling. In general, the problem with BSP trees is that polygons frequently have to be split as they are registered in the tree, and, for an n polygon scene, ....

Bruce F. Naylor. Partitioning tree image representation and generation from 3D geometric models. In Proceedings of Graphics Interface '92, pages 201--212, May 1992.

....step that identifies nearby large occluders for all viewpoints. Finally, a hierarchical visibility algorithm repeatedly applies the visibility test to determine the status of tree nodes in a spatial hierarchy. 1. 1 Related Work Given a 3D model and a viewpoint, exact visibility algorithms [12, 14] compute a description of the image in terms of visible polygon fragments. Once such a description is available, further processing can be restricted to involve only the visible portions of the scene. However, these techniques tend to be complex and hence difficult to use in interactive ....

Naylor, B. F. Partitioning Tree Image Representation and Generation from 3D geometric models. In Proc. Graphics Interface '92 (1992), pp. 201--211.

....2.5 Assemblage of Volume Primitives Various methods have been used to represent an object with structures of volume primitives. We will mention three widely different ones. 2.5. 1 BSP Trees Binary Space Partitioning (BSP) trees are structures made of the union and intersection of half spaces [9]. They are widely used not only as data structures for efficient spatial query and navigation, but also as modeling tools [10] For volumetric data, the strategy is to generate the BSP trees by fitting planes to the boundary points. This needs good heuristics to be effective. Once this is done, ....

Bruce F. Naylor, "Partitioning Tree Image Representation and Generation from 3D Geometric Models", Proceedings of Graphics Interface '92, May 1992, pp. 201--212.

No context found.

Bruce F. Naylor. Partitioning tree image representation and generation from 3D geometric models. In Proceedings of Graphics Interface '92, pages 201--212, May 1992.

No context found.

Naylor BF. Partitioning tree image representation and generation from 3D geometric models. Proceedings of Graphics Interface 1992. p. 201}12.

No context found.

Naylor, B., Partitioning tree image representation and generation from 3D geometric models, Proceedings of Graphics Interface '92, 201-212, may 1992.

No context found.

B. F. Naylor, "Partitioning tree image representation and generation from 3D geometric models", in Proceedings of Graphics Interface '92, (Vancouver), pp. 201--212, (May 1992).

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC