Yesterday on the main exhibition floor, SGI had presentations from various visual effect companies showing off their work. WETA happened to be sharing when I went and they were showing off MASSIVE, the Behavioural Intelligence software that is being used for crowd generation and the battle scenes in the Lord of the Rings.

The software looks very much likie Houdini's node view. There were hundreds of nodes that connected to create a master "brain".

They showed several examples of crowd generation that was brilliant. The basic process is they take motion capture data from real actors, then plug them into individual nodes in MASSIVE. The software make decisions IN REAL TIME of what the next behaviour will be based on how the user creates the brain.

They had an example of 2,000 characters (1,000 for each army) charging and going into battle. What they didn't predict was that if a character didn't see an enemy, he would run away. The "brain" was basically evolving on its own and making logical (or illogical depending how you look at it) decisions.

Another example included a warrior walking on uneven terrain. What was spectacular about this example was the walk cycle was updating in real time based on the steepness of the terrain so it was constantly changing to look like a natural gait.

Then they showed fully rendered orcs walking together based on this same approach and each had a distinctly different way of carrying their weapons and striding. In fact, the simulation showed how some of the orcs stride and drift to one side to give diversity to the scene.

What was really intriguing to hear was that WETA is creating battle scenes that will have 100,000 characters interacting with each other in battle...this includes fighting, clothing animation, blood, behaviour, and death. Eat your heart out ILM.