ETD

• after effect
• data driven
• ffmpeg
• parallel
• video automation

The main objective of this thesis is to produce video clips automatically in parallel from an input (XML/JSON) data and Adobe After Effects template. The project implementation is based on an open source plugin which partitions a single input job (XML/JSON) into a set of jobs and assigns each part to a set of distributed worker nodes in order to process (render) it with Adobe After Effects rendering engine. But since Adobe After Effects does not allow to output video when a job is rendered on multiple machines (render farm), the outputs of the above worker nodes are image sequences (also called still images). This C++ project is used as a sink to the above stream generator and is able to reconstruct videos in parallel and finally merge (concatenate) the reconstructed videos to generate the final high-quality output video resulting in a much better scalability, speedup and efficiency.