Wednesday, May 23rd, 2012 Posted by Jim Thacker

Videos: technical advances from Siggraph 2012

The Siggraph organisers have posted a cool preview video showing the highlights of the technical papers to be presented at Siggraph 2012.

A lot of this year’s research relates to fluid simulation, in one form or another: Naiad co-creator Robert Bridson contributes the ‘Ghost SPH scheme‘ – an improvement to conventional Smoothed Particle Hydrodynamics when it comes to simulating cohesion between a liquid and a solid object it is flowing over.

There is also a rather elegant new solution for simulating surface foam by treating individual bubbles as sites in a Voronoi diagram; and new work on smoke simulation co-authored by Scanline VFX‘s Nils Thuerey.

Tobias Pfaff, Nils Thuerey and Markus Gross’ work on using Lagrangian Vortex Sheets for animating fluids: an approach that can be used to create ultra-detailed animations of dense smoke plumes, among other things.

Another paper that caught our eye was one on using displacement maps on viscoelastic materials – mainly, we have to admit, because we like watching simulations of goo being dropped from a great height.

One piece of research that could have a much wider uptake, however, is Stanford University’s work on component-based shape synthesis. The video below shows a fleet of variant aircraft generated from a small set of ‘seed’ models – something that could easily find a home in building game asset libraries, for example.

Evangelos Kalogerakis, Siddhartha Chaudhuri, Daphne Koller and Vladlen Koltun’s work on component-based shape synthesis automatically generates new forms based on a small set of seed models.

Over in the field of 2D, there is an impressive new patch-based ‘image melding‘ system for synthesising plausible-looking new images from disparate source images. Several of the authors are from Adobe, and it’s not hard to see it making its way into something like Photoshop’s new Content-Aware tools.

But perhaps our favourite was HelpingHand, which automatically converts novices’ drawings into “the more confident strokes of a trained artist”. At this stage, it’s more a means of varying stroke width than it is of instantly converting your doodles to Michaelangelo drawings, but it’s still extremely cool.

All of this work represents early-stage research, so it may be a while before these technologies are integrated into studios’ proprietary tools, let alone off-the-shelf products – but it’s still a glimpse of the future.

You can see brief tasters of all of these papers and more in the highlight reel at the top of this post. Longer video demonstrations of some of the research can be found on ACM Siggraph’s YouTube channel.