Last TimeTodayRendering Massive ModelsImage-Based Rendering RevisitedBillboardsView Dependent BillboardsView Dependent TexturesImpostor ExampleTextured Clusters (Maciel and Shirley, I3D 1995)Clusters and Tree WalkClusters Results - ImageClusters Results – Top ViewTextured Clusters DiscussionUrban Scenery (Sillion, Drettakis and Bodelet, EG 1997)Impostor ConstructionUrban Impostors DetailsUrban Impostors - ResultsSlide 18Multi-Layered Impostors (Decoret, Schaufler, Sillion and Dorsey, EG 1999)Multi-Layer Impostors - ResultsSlide 21Coherent Layers (Lengyel and Snyder, SIGGRAPH 1997)Chicken Run (80 layers)Other ExamplesImage Caching (1) (Schaufler and Sturzlinger, EG 1996)Image Caching ExampleImage Caching (2) (Shade, Lischinski, Salesin, DeRose, Snyder, SIGGRAPH 1996)Layered Depth Images (Shade, Gortler, He and Szeliski, SIGGRAPH 1998)Open Problems in IBRNext Time03/14/05 © 2005 University of WisconsinLast Time•Image Based Rendering–Surface Light Fields–Architecture–Faces•Projects: By the end of this week03/14/05 © 2005 University of WisconsinToday•Image-based methods for real time rendering03/14/05 © 2005 University of WisconsinRendering Massive Models•“Images” for image-based rendering need not be images of real-things•IBR is an important technique for reducing the cost of rendering large models–Large these days means millions of polygons–Many models exist of this size: cities, architectural models, ships, factories, …•Level-of-detail techniques aim to reduce the cost of rendering by reducing the triangle count–Whole books on this topic03/14/05 © 2005 University of WisconsinImage-Based Rendering Revisited•Image-based rendering may be applied to the problem of rendering large databases–Advantage: Rendering cost depends mostly on the number of pixels in the images, not the number of objects those pixels represent–Disadvantage: Have to allow a wide range of viewer motion, which exacerbates typical IBR problems (cracks, stretching…)•Basic idea: Replace geometry with a few texture mapped polygons–Polygons and texture maps can be generated in a pre-process, or as required at run-time•Environment maps are an instance of this approach03/14/05 © 2005 University of WisconsinBillboards•A billboard is extreme LOD, reducing all the geometry to one or more textured polygons–The geometry is replaced by an image in the form of a texture•Issues in designing billboards–How are they generated?–How are they oriented with respect to the viewer?–How can they be improved?–None of the answers in this class•Also called sprites, but a sprite normally stays aligned parallel to the image plane03/14/05 © 2005 University of WisconsinView Dependent Billboards•What if the object is not rotationally symmetric?–Appearance should change from different viewing angles•This can still be done with billboards:–Compute multiple textures each corresponding to a different view–Keep polygon fixed but vary texture according to viewer direction–Best: Interpolate, with texture blending, between the two nearest views•Can use 3D textures and hardware texture filtering to achieve good results•Polygons are typically fixed in this approach, which restricts the viewing angles–Solution: Use more polygons each with a set of views associated with it03/14/05 © 2005 University of WisconsinView Dependent TexturesScreen shots from an Nvidia demo03/14/05 © 2005 University of WisconsinImpostor Example•Another methods uses slices from the original volume and blends themhttp://zeus.fri.uni-lj.si/~aleks/slicing-and-blending/03/14/05 © 2005 University of WisconsinTextured Clusters(Maciel and Shirley, I3D 1995)•Provide a range of representations for each object, and a benefit associated with each representation–Texture-mapped box is one representation, as is an average color box–Benefit may be view dependent – only texture one face•Use an octree to group objects into clusters–Representations for a cluster are a textured and average color box–Benefit is derived from benefit of children (maximum of children)•Deciding what to draw at run time is a bin-packing variation–Solve approximately•First pass computes cost, benefit and visibility, and sets initial model•Second pass starts at root, expands nodes in order of decreasing benefit–Not very scalable – looks at every representation of every object on each frame03/14/05 © 2005 University of WisconsinClusters and Tree Walk03/14/05 © 2005 University of WisconsinClusters Results - Image03/14/05 © 2005 University of WisconsinClusters Results – Top View03/14/05 © 2005 University of WisconsinTextured Clusters Discussion•Coined the term “impostor”•Benefit does not add: The benefit of two objects together may be much greater than their sum (eg a man and a gun)•Data-requirements may be very large–The appearance of a cluster can change greatly with viewing position•Shading highlights will change rapidly•Bigger clusters near the root contain many objects that may show significant disparity–Solution: Generate the necessary textures at run-time•The problem is easier for special cases…03/14/05 © 2005 University of WisconsinUrban Scenery(Sillion, Drettakis and Bodelet, EG 1997)•Dense cities have two useful properties when viewed from the ground:–The view location is highly constrained to be on streets–When standing on a street, the only distant views are through the ends of the street•Place an impostor at the ends of each street–Each impostor is a textured 3D mesh•At run-time, render the local model (the street and its buildings) and an impostor (showing everything else)•Additional space cost is linear in the number of streets•Rendering is much faster, because a relatively constant, small number of polygons are drawn for each frame03/14/05 © 2005 University of WisconsinImpostor Construction03/14/05 © 2005 University of WisconsinUrban Impostors Details•Impostors are generated in a pre-processing stage–Render the view from the center of each street out each end–Take the depth map, and triangulate it while attempting to preserve major discontinuities–Use the image as a texture for the triangle mesh•Remaining problems:–Transitions at intersections are problematic•At an intersection, the relevant impostors are most incorrect–3D mesh impostors stretch over what should be holes–More impostors per street are required for accurate occlusion