I’ve been implementing a design tool for creating an animation based on the real-world movement. In the previous projects, I had 1) basic sketching or importing image UI, 2) skeleton and joint detection. For the final project, I’m planning to implement two parts:
- Deforming hand-drawn pictures without distortion
- Capturing real-world body movements
For deforming hand-drawn characters, I’ll use As-Rigid-As-Possible algorithm discussed in this paper. Also, I’m thinking to capture the body movement with depth camera like Kinect or image recognition techniques like OpenCV.
The goals of the project is the followings
Minimum: If I cannot successfully use the Kinect or image recognitions to detect a skeleton or body movements, I’ll use a facial recognition instead to animate facial expression of the character.
Target: Animate imported human-shape character based on movements captured by Kinect or OpenCV
Stretch: Animate any kinds of shapes (not only human but also something like dog, chair, etc)