Opensteer is an open-source c++ library that provides a number of steering behaviours to drive autonomous agents. Created by Craig Reynolds of SCEA research (the original creator of “boids” in 1986, one of the first computer models to demonstrate flocking behaviour), it offers behaviours such as seek, flee, align, and object avoidance. It comes with a very easy-to-use demo framework that allows you to easily develop quick prototypes of ideas.

Definitely one of the best news/blog sites I’ve come across, We Make Money Not Art has more converage of the Digital Arts Festival Tokyo from last week. Among the coverage is a short write up of Toshio Iwai’s latest work, Morphovision, a new work exploring his theme of warping reality.

This latest installation gives the user the experience of his vision being warped, not through the use of a screen, but directly affecting a real object he’s looking at. This is done by apparently rapidly spinning an object and altering the timing and pattern of light hitting the object. As the light hitting the eyes is slightly delayed when reflected off certain parts of the model, different parts appear at slightly different rotatations, making the model appear melty and soft. Genius.

Follow this Link to a Japanese description.

Like something out of a science fiction film, these installations created by Sachiko Kodama of the Magnetic Fluid Art Project, are stunning and more than a little unnerving. Using a NASA developed fluid that reacts strongly to magnetic fields, they modify the surrounding magnetic field in response to sound input, to generate patterns that the fluid is bound to follow. The result is a series of stunning organic patterns, reacting to voice, music and ambient sounds, in a form that appears to be a ghibli-esque amorphous lifeform, that performs like a pet for the artists. Mesmerisingly creepy, I’m amazed that this hasnt been used more (although I’m guessing that the kit is probably quite pricey). Would be so great to connect to a webcam input device for an Abyss-style interaction with the blob.

(Found from the ever-amazing We Make Money Not Art)

Been having some fun making a little installation with Duncan that implements some simple computer vision techniques. The installation has a little snow particle simulation combined with webcam input to snow-ify people that happen to pass the window of where its going to be installed. Its been done before, most notably by the incredible folk over at Mine Control, but its been fun to make. It just uses the Vidcapture object to grab frames from any DirectShow device, combined with a Sobel filter to detect the edges, and then the particles move freely unless they drift into an area with sufficient gradient to make them stick. It doesnt have a bias for x or y gradients at the moment, so looks a little odd as snow sticks to vertical surfaces, but a bit more tweaking should getting it looking more realistic.