Computer Vision and graphics applied to digital video compositing and motion-design
Speaker: Alexandre Gauthier-Foichat, startup manager at Inria, INRIA Rhône-Alpes, Montbonnot, France
Abstract: The creation workflow of television films and advertising nowadays relies on many sophisticated pieces of software that apply computer vision and graphics to achieve animated imagery results that would otherwise be impossible without them. Many aspects of an image can be modified in such workflow: Live action camera capture, tracking, keying, grading… Moreover, it is possible to integrate seamlessly any virtual computer generated animated objects and characters.
The overall pipeline is complex and involves many expert users and different pieces of software. This presentation focuses on the taking and ending of compositing, which is a cornerstone of most production pipelines. We will show what technical challenges are involved and what the existing state of the art solutions are.
Keywords: Compositing, Real-time rendering, 2D/3D Tracking, Direct Acyclic Graphs, Caching, GPGPU, Multi-threading.
PAISS, Artificial Intelligence Summer School, 2-6 July 2018 , co-organizied by Inria and NAVER LABS Europe.