Autodesk has used GDC 09 to announce new versions of two of its primary middleware technologies.
HumanIK 4 updates the company's animation middleware, while Kynapse 6 offers an improved edition of the Autodesk artificial intelligence solution.
“Autodesk middleware products offer a solution to common runtime challenges, allowing development teams to concentrate on the creative work involved in authoring amazing new gameplay experiences,” said Marc Stevens, Autodesk vice president, games.
“For example, HumanIK helps to alleviate the burden of large clip libraries by enabling procedural motion adaption, which reduces the number of clips animators need to produce and maintain. With Kynapse, programmers can avoid the development time and costs involved in writing custom runtime solutions for universal AI issues, like spatial awareness.”
Human IK 4 introduces a user guide and a new set of examples, integrated with the Trinigy Vision game engine. It also improves knee response to ankle rotations.
Meanwhile, Kynapse 6 offers a range of updates as detailed in Autodesk's latest GDC 09 press statement:
• Remote debugger: The new remote debugging tool enables users to inspect the state of the game variables, with an interactive 3D view of the debug information. Users can also record and replay sequences, while retaining access to the data in the recorded files.
• New 3D pathfinding technology: The “flat” pathfinding technology introduced in this release offers simplified data generation workflows, more flexible runtime data streaming, and the ability to add new paths on-the-fly at runtime.
• Improved MMO support: Data resources and runtime services can now be shared across multiple Kynapse worlds, improving support for massively multiplayer on-line games.
• Enhanced dynamic 3D pathfinding: With improved CPU performance and granular control over memory consumption, characters can more easily avoid dynamic, movable obstacles.
• Hierarchical 3D pathfinding: Characters can now plan paths that span the entire breadth of huge maps, even when those maps cannot be loaded in their entirety. The path is first computed at a low level of detail, and then refined opportunistically when more detailed data becomes available, keeping memory usage within set limits.