Computer Vision News - September 2022

7 Synthesia with Matthias Nießner Synthesia has a strong founding team, including Matthias, Lourdes Agapito , Victor Riparbelli , and Steffen Tjerrild . Matthias tells us that Victor and Steffen essentially run the company while he brings the academic side. “ There’s a big difference between publishing a paper and making a viable and sustainable company, ” he adds. “ Our people are our secret sauce. We have a nice combination of academics who do the crazy tech, the latest generative deep learning models, and the more pragmatic, like what will work in practice? Having that communication on a very level playing field is super important. This synergy works very well for us, and I think we have the perfect team for it. ” Synthetic media , in general, has had some has hired three interns, funded PhD students, and aims to publish papers at upcoming top- tier computer vision events. This moonshot research may not end up in the product immediately but is ultimately intended to advance the state of the art. “ We feel that video-generative models, GANs, and diffusion models are still not at the level where very generic videos can be synthesized, ” Matthias points out. “ There’s still a gap in what research can do. Work on the latest generative models and implicit representations is something you will see much more of from us. ”