CS 563 Advanced Topics in Computer Graphics View

CS 563 Advanced Topics in Computer Graphics View Interpolation and Image Warping by Brad Goodwin Images in this presentation are used WITHOUT permission

Over View § General Imaged-Based Rendering § Interpolation § Plenoptic Function § Layered Depth Image (LDI)

Introduction § Image Based Rendering (IBR) § § § § Composed of photometric observations Mix of fields (photogrammetry, vision, graphics) Texture mapping Environment mapping Realistic surface models Uses from virtual reality to video games Just render the 3 D scene? Judge results? § Different types of rendering using different amounts of geometry

Interpolation § Morphing § interpolating texture map and shape § Generation of a new image is independent of scene complexity § Morph adjacent images to view between § based on viewpoints being closely spaced § Uses camera position, orientation and range to deteremine pixel by pixel § Images pre computed and stored as morph maps

About this method § Method can be applied to natural images § Only synthetic were tested with this paper § Of course this paper was in ’ 93 so hopefully someone’s tested them by now § Only accurately supports view independent shading § Others could be used on maps but they are discussed

Types of Images § Can be done with natural or sythetic images § Sythetic § easy to get the range and camera data § Natural § Use ranging camera § Computed by photogrammetry or artist

General Setup § Morphing can interpolate different parameters § § Camera position Viewing angle Direction of view Hierarchical object transformation § Find correspondence of images § Images arranged in graph structure

Find correspondence § Usually done by animator § This method § Form of forward mapping § uses camera and range to do it § Cross dissolving pixels(not view-independent) § Done for each source image § Quadtree compression § Move groups of pixels § Scene moves opposite camera § Offset vectors for each pixel (“morph map”) § Small change more accurate when interpolated

§ Sampled every 20 pixels Offset vectors

Overlaps and holes § Overlaps § Local image contraction - several samples move to the same pixel in interpolated image § Perpendicular to oblique § Holes § Show when mapping source to destination § Background color § Interpolate four corners of the pixel instead of center (filling and filtering) § Interpolate adjacent offset vectors § Or if part seen in interpolated but not source

Block Compression § Pixels ten to move together so block compression algorithm is used to compress morph map. § Related to image depth complexity § High complexity low compression ratio

View independent Priority § Established to determine points that are viewable § Pixels are ordered from back to front based on Zcoordinates established in morph map § Eliminates need for interpolating the Z-coordinates of every pixel and updating the Z-buffer in the interpolation process.

Applications § Virtual Reality § Motion blur § Uses super-sampling of many images computationally which is expensive thus inefficient § Reduce cost of computing a shadow map § Only for point light sources § Create 3 D primitives without creating 3 D primitives

Plenoptic Modeling § The Plenoptic function § Latin root plenus – complete or full optic - pertaining to vision § Parameterized function for describing everything that is visible from a given point in space § Used as a taxonomy to evaluate low-level vision § Adelson and Bergen postulate “…all the basic visual measurements can be considered to characterize local change along one or tow dimensions of a single function that describes the sructure of the information in the light impinging on an observer. ”

Parameters § azimuth and elevation angle

Plenoptic § Set of all possible environment maps for a given scene § Specify point and range for some constant t § A complete sample can be defined as a full spherical map

Plenoptic Modeling § Claimed that all image-based rendering approaches are just attempts to create a plenoptic function with just a sampling of it § Set up is the same as most approaches § Set of reference images which are warped to create instances of the scene from arbitrary view points

Sample Representation § Unit sphere § Hard to store on a computer § Example of all distorted maps § Six planar projections of a cube § Easy to store § 90 degree face requires expensive lens system to avoid distortion § Oversampling in corners § Have to choose Cylindrical § Easily unrolled § Finite height : problems with boundary conditions § No end caps

Aquiring Cylindrical Projections § Get the projections is simple § Tripod that can continuously pan § Ideally camera’s panning motion should be exact center of tripod § When panning objects are far away slight misalignment is tolerated § Panning takes place entirely on the x-z plane § Both images should have points within each other.

§ Find the projection of the output camera on input cameras image plane § That is the intersection of the line joining the two camera locations with the input camera’s image plane § Line joining the two cameras is the epipolar line § Intersection with the image plane is the epipolar point

§ Map image point to output cylinder § Same techique for comparing points used with face mapping from last week

Layered Depth Images § Paper presents some methods to render multiple frames per second on a PC § Sprites – are texture maps or images with alphas (transparent pixels) rendered onto planar surfaces § One method warps Sprits with Depth § Warps depth values and uses this information to add parallax correction to a standard sprite renderer § LDI § Single input camera § Contains multiple pixels along each line of sight § Size of representation grows linearly with the depth complexity of the scene § Uses Mc. Millan’s warp odering algorithm because data is represented in a single image coordinate system.

References § § § § Chen S E and Williams L, "View Interpolation for Image Synthesis", Proc. ACM SIGGRAPH '93 Mc. Millan L, and Bishop, "Plenoptic Modeling: An Image-based Rendering System", Proc. ACM SIGGRAPH '95 Shade, Gortler, He and Szeliski, "Layered-Depth Images", Proc. ACM SIGGRAPH '98 Mc. Millan L. and Gortler S, "Applications of Computer Vision to Computer Graphics: Image-Based Rendering - A New Interface Between Computer Vision and Computer Graphics, ACM SIGGRAPH Computer Graphics Newsletter, vol 33, No. 4, November 1999 Shum, Heung-Yeung and Kang, Sing Bing, A Review of Imagebased Rendering Techniques, Microsoft Research Watt, 3 D Graphics 2000, Image-based rendering and phtomodeling (Ch 16) http: //www. widearea. co. uk/designer/anti. html http: //www. dai. ed. ac. uk/CVonline/LOCAL_COPIES/EPSRC_SSAZ/n ode 18. html http: //www. cs. northwestern. edu/~watsonb/school/teaching/395. 2 /presentations/14

Questions? ? ? ?