Student presentations - Dec 6 - double header 2:00-4:30
=====================
~7.5 minutes total per presentation, details from TA
   5 minute talk
   2.5 minutes questions

Outline:
* Motivation and Problem
* Approach
* Results
* Discussion

Topics for 2nd exam - Dec 1 - Open book
===================
Comprehensive exam, but focus on:
  - parametric curves [covered by Assn3]
  - parametric surfaces
  - subdivision surfaces & curves
  - particle systems [covered by Assn3]
  - projections and Z-buffer
     . perspective transformation


Recap of what we've learned
===========================
* outline of course
* things you can learn about:
   . skills / methods of thinking/inquiry:
       proofs, mathematical analysis, coding (easy part)
       synthesis of complex systems (hard part),
       research (problem identification and solution generation;
       formalization of problems),
       critical thinking
   . facts
   . problem domain:
      - basic problem to be solved - physics, math in this case
      - tradeoffs: e.g. performance vs. fidelity
      - in our case: importance of *approximation*!
   . "best practices" -- proven approaches to solving problem
      . inherently domain-specific
      . but illuminate usefulness of basic CS ideas and makes them concrete
      . importance of learning through induction!
      . actual exposure to complexity and uncertainty

* broad coverage of many topics:
   . continuous-domain math:
       . surface & curve representations
       . anti-aliasing
       . shading
       . sampling and numerical integration (transition to discrete domain)
       . 3D translation and rotation
   . physics
       . light transport
       . simple mechanics (particle systems)
       . material properties and interaction with light
   . basic data structures and algorithms
       . spatial data structures for ray tracing acceleration
       . visibility / hidden-surface problem
   . fact dump
      - human perception
      - display devices
   . best practices - how application problem is solved in practice.
          algorithms; data structures; factoring of problem;
          system organization; etc.
	  This is 50% of the heart of "applications" areas.
	  But beware of changes as time progresses!
	  Relatively speaking, we have de-emphasized this material in
	  this course!
   . performance optimization and system issues
      - Z-buffer algorithm and architecture
      - in large part an example of "best practices"


What we have not covered in this class
======================================
Rendering:
  * Line rasterization
  * Triangle rasterization
  * Triangle clipping
  * Texture filtering, in detail
  * Procedural texturing -- noise, etc.
  * Geometric level-of-detail management; simplification algorithms
  * Z-buffer visibility culling algorithms
  * Z-buffer hacks (environment mapping; cube maps; ...)
  * REYES
  * other visibility algorithms (esp. old ones)
  * Global illumination in depth (this is another class)
  * Z-buffer and graphics hardware in depth (this is another class)
  * non-photorealistic rendering
  * volume rendering
Modeling:
  * CSG
  * Image-based modelling: Acquiring and redisplaying real-world images
Visualization:
  * Conveying information to user
     . scientific visualization
     . medical visualization
     . "information" visualization
Physics:
  * Collision detection (in detail)
  * Fluids, flame, etc... (this is a whole course)
Computational geometry:
  * Representation of meshes; operations on them; etc.
Animation:
  * Lots of recent work in "steerable" animation
User interfaces:
  * Mostly a 2D topic
  * Window systems
Overall:
  * Could go into substantially more depth for every topic we
    examined.

Graphics researchers at UT
==========================
* Myself:
    - Systems-ish
    - Focus: real-time graphics; graphics architectures;
             single-chip parallel systems.
    - Questions:
        . Can we use global illumination algorithms, or approximations
	  thereof, in real-time?  What algorithms and architectures
	  are appropriate?
	. What should the architecture and programming models for
	  single-chip parallel architectures be?
* Don Fussell
    - Also systems focused; but past work on light fields, fractals, ...
* Chandrajit Bajaj
    - More mathematical
    - Surface and volume representations;
        modelling and visualization of protein docking, etc.
* Bruce Naylor (adjunct)
    - BSP trees; audio rendering
* Kelly Gaither (TACC research associate)
    - scientific visualization algorithms and systems (esp. volume rendering)

Research resources
==================
* SIGGRAPH
* Graphics Hardware
* Eurographics Rendering Conference
* Interactive 3D Conference
* IEEE Visualization Conference