This paper combines two well-known algorithms for fluid simulation into a novel, robust and stable fluid simulation algorithm. The paper performs a strong analysis of the pro and cons of existing techniques which leads to the novel algorithm combining the best of both techniques, without their drawbacks. The results are excellent and should lead to practical benefits for most fluid simulations. The presentation at the conference was exceptionally clear, and all members of the committee agreed in awarding this contribution with the 2014 Günter Enderle best paper award.

This paper introduces a very efficient parallel implementation for L-system grammar evaluation. This innovative algorithm will likely have a strong practical impact as it changes the way L-systems can be used; from off-line generation to run-time evaluation during user exploration of a scene.

This paper introduces a novel approach to unsupervised part classification within a collection of objects, using the arrangement of parts as a similarity measure instead of their actual geometry. The approach produces impressive results on a variety of object collections. Using higher-level information such as the spatial relationships between parts to help the analysis of large collections is innovative and likely to inspire future work.