fxpodcast #349 Move AI

Move.ai enables performance capture and movement between the physical and the digital world, using patented artificial intelligence software to extract high-fidelity motion from video. Move.ai’s artificial intelligence, computer vision, and machine learning software can capture cinematic quality motion data anywhere from video and even a set of iPhones, – without suits, markers, special lighting or MoCap cameras.

In this episode, we talk to Anthony Ganjou, co-founder of Move.ai. In the episode, we discuss how to get the best out of their Machine Learning software as well as some of the future directions the company will be exploring such as cloth and face performance capture. The discussion zeros on the company’s new multiple iPhone capture system ($365).

The guys also discuss tests done by EA and presented at SIGGRAPH. Below is the 30 mins talk from August 2022’s SIGGRAPH.

The EA tests compared an optical system and an inertial system with a markerless Move.ai approach.

The results are very impressive and show just how far machine learning solutions have come.

The discussion talks about the multiple iPhone capture approach that is now coming out of Beta and also touches on the advances in finger tracking and new spine approximations. The spine is often estimated with just three points in many MoCap rigs and this does not adequately represent the curvature of a real spine. Future research is expected to also solve for cloth and facial expressions.

 

Leave a Reply