Q. Chen, F. Tian, H.S. Seah, J. Qiu, and K. Melikhov (Singapore)
Computer-assisted animation, motion estimation, hierar chical processing
Our research focuses on computer-generated inbetween ing, which aims to reduce the workload of manual draw ing in traditional cartoon production. Our previous method, Modified Feature-Based Matching Algorithm, MFBA, es timates the motion of the pixels in the two keyframes. In betweens are generated by interpolating the motion vec tors. However, the estimation may not be accurate when the change between the keyframes is drastic. In this paper, the influence of the initial value is examined in the mo tion estimation of MFBA. Consequently a method is pro posed to evaluate the approximate motion of the character and specify it as the initial value in the motion estimation. The results show smoother motion in the inbetweens.
Important Links:
Go Back
