Many interactive applications require realistic ,high-quality character animation.This demand for realistic motion will almost certainly increase. However, motion usually competes for memory and other resources. Thus,Heightened realism in motion cannot simply come from more data. Motion will probably need to be generated from a small dataset and modified to enrich varibility.
Motion belnding ([Kovar and Gleicher 2004][ Rose et al. 1998 ]) and motion warping([ Witkin and Popovic 1995]) are commonly used motion editing techniques .Another commonly used method rearranges motion frames to produce novel motion sequences.Motion graphs([ Kovar et al. 2002a],[Lee at al.2002],[Arikan and Forsyth 2002]) encode which rearrangements are legal.
論文的這一段，非常好的闡釋了我們的運(yùn)動編輯和運(yùn)動圖的必要性。Mark，困擾的問題正在一點點解開。。