Google’s latest Project Starline prototype uses AI to make 3D images of people

Today at I/O, Google announced the latest prototype of Project Starline, its 3D teleconferencing video booth. On the face of it, the biggest difference a dramatic reduction in hardware. The earlier model, which it showed off late last year, relied on several cameras to capture speakers from all angles. This time out, however, the system only has a “few” of them, leaning on AI and ML to generate a convincing three-dimensional image of its subject.

According to Google, the system has gone “from the size of a restaurant booth to a flat-screen TV,” courtesy of these hardware updates. It notes in a blog post,

Our earlier Project Starline prototypes took up an entire room, requiring complex hardware such as infrared light emitters and special cameras to create a live 3D model of the person you were talking to. While the results were impressive, the size and complexity of the system made it challenging to bring to many of today’s offices.

Google has already begun testing the latest prototype with a number of high profile partners, including T-Mobile, Salesforce and WeWork.

“Trust and innovation are at the heart of Salesforce’s values,” says Salesforce SVP Andy White. “Project Starline has the potential to help create authentic and immersive connections that foster deeper relationships with both our employees and customers, enhance trust and transparency, and drive productivity and efficiency.”

The news arrives on the heels of Logitech’s Project Ghost teleconferencing booth — though that device offers a more two-dimensional experience in a much larger form factor. Both products arrive amid a transformed workforce, when working from home has become the rule – rather than exception – for many companies.

No specifics on timeline – or price – for the new system.

Read more about Google I/O 2023 on TechCrunch