Motion Merging

I thought I'd take the time to post a quick overview of the work I did for my thesis (for my M.S. in Computer Science). My thesis work involved working with two independent animation sequences and merging them together in order to attempt to aid in alleviating production schedules. The idea of “motion merging” comes in to place as an alternative to such time inefficient tasks as key framing and graph editing. Merging two independent motions simply means that two animated sequences are supplied and my algorithm takes those sequences and merges them in to one unique sequence.

Several solutions have been discovered towards this "in-betweening" effort including but not limited to key framing, linear interpolation, motion re-targeting, and even manually animating every frame. However, each of these solutions are costly and time consuming. Using quaternions and a form of linear interpolation, a mathematical approach was used to “blend” two independent animated sequences and aid in streamlining this process.

Using the basic guy rig found at The 11 Second Club, I was able to test my algorithm and verify it's success. Following you'll see the videos demonstrating the "in-betweening". The first video uses the basic guy rig where I 'motion merge' the basic guy from a standing position to a sitting position. The second video uses two completely different objects (a cone and a cylinder) where the first object is animated to merge in to the second object's position.

Basic Guy

Cone-to-Cylinder

There you have it! Motion merging was a really fun topic to work on. I'll try and post more details on how the algorithm works at a later date.