Google Seurat is an Amazing Baby Step for VR

Steven Levy wrote a very good deep dive into Google’s mixed reality efforts for Backchannel a few weeks ago. Published on the eve of the Google I/O conference, Levy’s post centers on Clay Bavor, Google’s VP of VR and AR and one of Silicon Valley’s biggest mixed reality cheerleaders.

Much of the post is centered around Bavor’s call to slow the wheels on the VR hype train. Bavor – and his contemporaries, Facebook’s Mark Zuckerberg among them – sees virtual reality as a long-term play, with truly life altering experiences most likely still a decade away from consumers’ reach. That said, Levy also detailed the big AR/VR announcements coming out of I/O, including forthcoming standalone Daydream headsets and Seurat, a new graphics optimization technology:

Google has also done some interesting graphics work to make up for not having a powerful computer driving the simulations. Its new Seurat technology uses a computational sleight of hand that dramatically cuts the rendering time of certain objects from hours to milliseconds. Indeed, when I tested this out earlier this week, the Rogue One scene I was plunged into seemed as dense as a high-end VR experience. No wonder: Google’s engineers told me that the original hi-res scene, as rendered by Lucasfilm, consisted of 50 million triangles, an order of magnitude more than even a powerful PC can handle. Seurat’s optimization tricks faithfully create the scene with only 70 thousand triangles, so it can easily run on the mobile Daydream platform in its full 3D glory.

Taken on its own, Seurat sounds like a significant feat of engineering with near-term potential for optimizing graphics beyond VR, as well. But in the context of the long, slow march towards true immersive reality, it sounds more like a baby step (if a very impressive and necessary one). And it’s just one of many to be taken over the next several years.