CS 497, Fall 2000
<br>
Prof. Yizhou Yu
<br>
<br>

Image-based Modeling and Rendering
<br>
<br>

Section YZY, #08826
<br>
3:30 – 4:45pm TuTh, 241 Everitt Lab.
<br>
<br>

Description: Image-based acquisition, reconstruction, representation, 
manipulation and display of graphical objects and scenes.
<br>
<br>

Prerequisite: CS318 computer graphics. Some knowledge about computer vision will be helpful, 
but not required.  Credit:  3/4 or 1 unit.
<br>
<br>

<p>
Image based modeling and rendering is a fairly new research trend in computer graphics that 
reconstructs and renders complex 3D scenes from a set of input images. It has considerably 
advanced the level of photorealism and performance of graphics. The use of images as input 
to this new modeling and rendering framework allows to capture far more complex 3D objects 
(i.e. highly complex sculptures) than can be achieved reasonably by traditional geometric 
modeling. Furthermore, the use of real photographs also enables to capture real-world lighting
information (i.e. unlimited number of light sources, and complex reflectance properties) that is 
almost impossible to model artificially.  While image based modeling allows complex scene and 
lighting  reconstruction, image based rendering provides a new and efficient  rendering paradigm.
Compared to traditional geometric rendering where the display performance and frame update 
rate is limited by the geometric complexity, number of polygons of the input, the performance 
of image based rendering is solely determined by the size and resolution of the output image and 
independent of the scene complexity.
</p>

Topics on image-based rendering:
<br>
1) Plenoptic function
<br>
2) Panoramic mosaic
<br>
3) Concentric mosaic
<br>
4) Light field rendering and Lumigraph
<br>
5) 3D morphing
<br>
6) Layered depth images
<br>
7) Texture synthesis and mapping
<br>
8) 3D texture
<br>
9) image-based re-lighting
<br>
<br>

Topics on image-based modeling:
<br>
1) Stereo
<br>
2) Shape from shading and photometric stereo
<br>
3) Hybrid approach
<br>
4) Voxel coloring and space carving
<br>
5) Laser scanning
<br>
<br>

Classes have mixed lectures and paper presentations.
<br>
Grading is based on class attendance, paper presentation and an optional term project.