David Kirk and Douglas Voorhies. The rendering architecture of the dn10000vs. In Proceedings of the 17th annual conference on Computer graphics and interactive techniques, pages 299--307. ACM Press, 1990. 23  Home/Search   Document Not in Database   Summary   Related Articles   Check

....curve and are therefore different from the interpolation approach described in this paper. Bishop and Weimer [Bis86] used a Taylor series expansion of the Phong equation in order to obtain a second order approximation. Seiler [Sei98] and Lee and Jen [Lee01] elaborated an idea by Kirk and Voorhies [Kir90], which is based on the fact that a quadratic shading surface can be determined from six sample points over the polygon, i.e. the intensity at the vertices and the edge midpoints) 3. REFORMULATION OF THE SHADING COMPUTATION Shoemake [Sho85] showed how spherical linear interpolation (slerp) ....

D. Kirk, D. Voorhies, The Rendering Architecture of the DN10000VS, Computer Graphics vol. 24, pp. 299-307, August 1990.

....raytraced rendering by efficiently answering queries about rays propagating through ordered sets of parallelepipedal cells. To our knowledge, these ray propagation techniques have not been used in interactive display systems. Given the wide availability of fast polygon rendering hardware [3, 14], it seems reasonable to search for simpler, faster algorithms which may overestimate the set of visible polygons, computing a superset of the true answer. Graphics hardware can then solve the 1 eye obstacle invisible object Figure 1: Cone octree culling: the boxed object is reported visible. ....

David Kirk and Douglas Voorhies. The rendering architecture of the DN10000VS. Computer Graphics (Proc. Siggraph '90), 24(4):299--307, 1990.

....amounts of money and man power to develop custom machines that will be obsolete in a few months time. We propose POMP as a non specialized architecture (with no hardware dedicated to any special algorithm) which is a step beyond the other alternatives (partially non specialized) proposed in [KV90, FPE 89] We have to balance the loss of power due to this non specialization by a massively parallel approach (up to 256 32 bit processors, in fact 8,192 bits of data paths) This massively parallel organization prohibits the organization in a multiprocessor with shared memory. Each ....

David Kirk and Douglas Voorhies. The rendering architecture of the dn10000vs. In Computer Graphics (SIGGRAPH '90), pages 299--307. Association for Computing Machinery, Aout 1990. Volume 24, Number 4.

....for each pixel on image a has been computed analytically. Images b, c, and d use first, second, and third order interpolation respectively. Since most hardware does not support higher order interpolation, this technique has mostly been used for off line rendering. Kirk and Voorhies [6] demonstrated second order color interpolation on the Apollo DN10000VS using the quadratic interpolation hardware built into that machine. To our knowledge, no one has demonstrated third order interpolation on a hardware based renderer. In this paper we demonstrate two ways to perform ....

....can also lead to increased frame rates because it takes fewer primitives, albeit more expensive ones, to equivalently represent the model. Quadratic Interpolation in Screen Space Two graphics workstations contain hardware for quadratic interpolation, Pixel Planes 5 [4] and the Apollo DN10000VS [6]. We have implemented second order color interpolated triangles as a basic primitive on Pixel Planes 5. At each pixel, the interpolation hardware computes the following expression Ax 2 Bxy Cy 2 Dx Ey F where x and y are the pixel coordinates in screen space, and A through F are the ....

Kirk, D., and D. Voorhies, The Rendering Architecture of the DN10000VS, Siggraph '90, 24:4, Aug. 1990, 299-308.

....and interpolated by the shading hardware before indexing into the texture table. If t varies linearly within each tetrahedron, the product l varies quadratically inside each projected triangle. Quadratic interpolation of texture coordinates was implemented in hardware on the Apollo DN10000VS [45]. Crawfis, Max and Becker [22] use a 2D texture table with t and l, and put 1 exp( l) in the table. Wilhelms and Van Gelder [102] extended Shirley and Tuchman s algorithm to generate polygons for a whole regular grid cell without the need to subdivide it into tetrahedra. This reduces the number ....

....added additional features that allow the user to watch moving puffs of smoke, control the time propagation of the smoke, and combine opaque geometry with the smoke. A general polyhedra sort [87] can be added and the smoke can be colored by a separate scalar variable. Meta balls [71] blobbies [45], or soft objects [69] are flexible algorithms for modeling fuzzy phenomena. A density volume is created by placing generating points in the volume s space. A basis function is centered at each of these points, contributing to the 25 overall density function throughout the space. Gaussian ....

Kirk, D. and D. Voorhies, The Rendering Architecture of the DN10000VS. Computer Graphics, 1990. 24(4): p. 299-307.

....amounts of money and man power to develop custom machines that will be obsolete in a few months time. We propose POMP as a non specialized architecture (with no hardware dedicated to any special algorithm) which is a step beyond the other alternatives (partially non specialized) proposed in [KV90, FPE 89] We have to balance the loss of power due to this non specialization by a massively parallel approach (up to 256 32 bit processors, in fact 8,192 bits of data paths) This massively parallel organization prohibits the organization in a multiprocessor with shared memory. Each ....

David Kirk and Douglas Voorhies. The rendering architecture of the dn10000vs. In Computer Graphics (SIGGRAPH '90), pages 299--307. Association for Computing Machinery, Aout 1990. Volume 24, Number 4.

....raytraced rendering by efficiently answering queries about rays propagating through ordered sets of parallelepipedal cells. To our knowledge, these ray propagation techniques have not been used in interactive display systems. Given the wide availability of fast polygon rendering hardware [3, 14], it seems reasonable to search for simpler, faster algorithms which may overestimate the set of visible polygons, computing a superset of the true answer. Graphics hardware can then solve the eye obstacle invisible object Figure 1: Cone octree culling: the boxed object is reported visible. ....

David Kirk and Douglas Voorhies. The rendering architecture of the DN10000VS. Computer Graphics (Proc. Siggraph '90), 24(4):299--307, 1990.

No context found.

David Kirk and Douglas Voorhies. The rendering architecture of the dn10000vs. In Proceedings of the 17th annual conference on Computer graphics and interactive techniques, pages 299--307. ACM Press, 1990. 23

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