In this paper, we present an algorithm for motion synthesis. We develop this algorithm as a part of a large crowd simulation project which simulates the motion of the people by using a large database of motion captured data. Two main features of this synthesis algorithm include generating motion along a given path with the speed matched and adapting the motion to the scaled skeleton. In some previous works, the similar algorithm are offline algorithm, memory consuming and possibly need to search neighbor time frames back and forth. In order to develop an online algorithm to be used in the large crowd simulation with small memory footprint, we develop this alternative solution.
|Title of host publication||Artificial Intelligence Techniques for Computer Graphics|
|Number of pages||13|
|Publication status||Published - 14 Nov 2008|
|Name||Studies in Computational Intelligence|