作者: Raquel Urtasun , Pascal Glardon , Ronan Boulic , Daniel Thalmann , Pascal Fua
DOI: 10.1111/J.1467-8659.2004.00809.X
关键词:
摘要: Representing motions as linear sums of principal components has become a widely accepted animation technique. While powerful, the simplest version this approach is not particularly well suited to modeling specific style an individual whose motion had yet been recorded when building database: it would take expert adjust PCA weights obtain that indistinguishable from his. Consequently, realism required, current practice perform full capture session each time new person must be considered. In paper, we extend so requirement can drastically reduced: for whole classes cyclic and noncyclic such walking, running or jumping, enough observe newcomer moving only once at particular speed jumping distance using either optical system simple pair synchronized video cameras. This one observation used compute set component best approximates extrapolate in real-time realistic animations same walking different speeds,