Class Projects

1
Class Projects

• The aim is to make some research contribution to
  the field of rendering
• Research doesnt always go as planned
• You will not be penalized if a reasonable initial
  hypothesis goes bad
• Try to structure your project with intermediate
  goals
• Work in pairs, or alone if you really want to

2
Time-line

• Select a project and have it approved by Nov 5,
  preferably sooner
• Progress reports (in class presentations) on Nov
  20
• Summary of work so far
• May include literature review, planned approach
• Final presentations on Dec 13
• In class presentations of you results (or
  anticipated)
• Final demonstrations and papers before Dec 20
• Demo if applicable
• 4-6 page paper

3
Other Stuff

• The project will form about 70 of your overall
  grade for the class
• Each sub-task (progress report, presentation etc)
  will contribute something
• Use whatever equipment you like
• I might be able to address special needs
• I can probably arrange access to various things
• The ultimate goal is SIGGRAPH papers

4
Metropolis Sampling for Extended Form Factors

• The Metropolis light transport algorithm has the
  disadvantage that it finely samples diffuse
  illumination.
• Its strength is in finding and sampling difficult
  paths.
• Apply Metropolis in a situation where its
  strengths are used to best advantage, while
  avoiding its weaknesses.
• One such application is in computing extended
  form factors (and also caustic photon maps).

5
Benchmarking Terrain LOD Algorithms

• Terrain level of detail is the problem of
  rendering terrain at minimal cost with low error
• There are several algorithms available for
  terrain LOD
• There are, however, no established benchmark
  terrains for comparing algorithms
• Implement the algorithms, design the benchmarks,
  and run the tests

6
Automatic Landscape Generation

• There are ways to generate terrain with
  particular random properties
• There are not many ways to automatically populate
  the terrain (with trees, houses, roads, grass,
  etc)
• For large terrains it may be possible to generate
  the landscape on demand, so that it need not be
  stored in full detail.

7
Greater Coherence in NPR Animations

• As is frequently stated, coherence is a problem
  in NPR animations
• Something better can surely be done, possibly by
  drawing on better probabilistic algorithms for
  placing strokes.

8
Cartoon Physics Rendering

• Cartoonists stretch and squish a ball as it
  bounces
• Generate such animations automatically
• A sketch at SIGGRAPH this year suggested one
  possible approach

9
NPR Contrast Manipulation

• Real paints and canvasses have real limitations
  on the range of brightnesses and colors they can
  represent
• Artists manipulate the light and shade in a
  composition in order to generate the perception
  of greater contrast or brightness is really
  present
• Vermeer is an excellent example he did it all
  the time.
• Technical illustration manuals also suggest
  enhancing the contrast across edges
• Try to develop a way to capture this effect.

10
Tone Reproduction

• Physically based rendering algorithms create
  accurate luminance values that must be visualized
• Monitors and papers don't have high dynamic range
• Cant reproduce the required luminance values
  directly
• Employ a Tone Reproduction Algorithm to squash
  the range of brightnesses down
• To something that can be reproduced
• Attempt to generate the same visual impact as the
  accurate brightness values.
• Create a new luminance visualization algorithm
• This problem is strongly related to the previous
  one