As I speak to Tong Pow over Google Meet, he appears as a cartoon medieval-helmet-clad cat that talks and bobs its head, mirroring Pow’s movement behind the webcam. The visor may be hiding a Cheshire cat grin because Pow has raised $6.5 million for his San Francisco-based startup Hologram Labs, just seven months after launching.
NFT avatars have become a common sight on social media platforms, representing their owners’ identities and status. But these blockchain-verified jpegs have limitations; they are static and don’t reflect changes in one’s physical self. Hologram wants to change that by bringing NFT avatars to life.
Pow, who worked at 0x Labs, a decentralized exchange protocol builder, and his co-founder Hongzi Mao, a computer science Ph.D. from MIT, have created a desktop tool that uses machine learning algorithms to turn still avatars into motion-tracked ones. Available as a Chrome extension, Hologram swaps out Pow’s real video feed with the cat avatar — which is part of an NFT collection called Cool Cats.
“Think of it as a machine learning model compressed and deployed to the browser. The technology has only been achievable in the last few years, with the speed and efficiency to run in real time,” explains Pow, adding that the AI model is trained using only publicly available data.
Once the model is deployed to the browser, it then feeds the face tracking points captured by the camera into the avatar, which enacts the movements live.
“To create motion-tracked characters, it requires either high-cost hardware for in-depth tracking or certain platforms that do not have the best user experience, so our angle is really creating the tooling and application for users to become virtual characters with one click,” the founder says.
Topics
