1. We propose a fully automated system that
constructs a cinematically expressive movie in real-time
from a flow of low-level narrative elements generated as the
environment evolves.

2. Traditional methods only look at isolated aspects of cinematography and do not allow for style variations.

3. Select narrative elements with a narrative engine, compose selected elements with analysis of visibility of key targets into director volumes, use a customizeable set of filters to select appropriate volumes, compute transitions between volumes.

4. This paper lets you automate all the cinematography components in real time while allowing for differences in director styling.

5. Slideshow: http://sites.google.com/site/christophelino/resources/2010-AReal-timeCinematographySystemforInteractive3DEnvironments-SCA-pres.pdf?attredirects=0
Video: http://sites.google.com/site/christophelino/resources/02_1984-example1-dynamicity.wmv?attredirects=0

1. We present a technique for learning clothing models that enables
the simultaneous animation of thousands of detailed garments in
real-time. [first sentence of abstract]

2. Modern cloth simulation is really nice, but expensive. Would like a compromise with low dimensional models but keep expressiveness of cloth discretization.

3. Observe cloth behavior under physical simulation to learn a simpler (non-physically based) proxy model. This model will generally behave well and is much less computationally expensive.

4. This paper lets you realistically simulate lots of cloth at the same time. Claimed naive approach (there is room for improvement) gave 3x faster result than Maya Cloth, rendered 170 real time clothing articles, or 1000 without rendering.

5. Reference Video: http://drp.disneyresearch.com/projects/cloth/videoNew.mp4