| Summary: | 碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 96 === Making computer animation has always been a complicated and time consuming job. Animation is achieved by first constructing a skeleton on the given mesh, and then by adjusting the skeleton to generate new poses.
In this thesis, we study how to make source models’ poses be used repeatedly. By transferring source model’s deformation gradient to target model, this simple method can copy source model’s pose to target model. It makes target model’s pose like source model’s pose. However, directly transfer source model’s deformation gradient to target model, the target model’s local detail deformation is not often ideal. So the proposed algorithm considers additional example models and gets some meaningful information from example model. Using the information, the detail deformation of target model likes example model’s detail. It can make the results look more perfect. The method generates target model’s pose that will save a lot of human’s labor effort and time.
|
|---|
