Varjo Teleport Lets You Capture Any Scene With An iPhone And View It Photorealistically In PC VR

Varjo Teleport lets you easily capture any scene with an iPhone and move around it in PC VR with photorealistic graphics.

Like the Gracia platform we reported on yesterday, Varjo Teleport is possible thanks to Gaussian splatting, a relatively new neural technique for rendering 3D volumes by representing the scene as a collection of overlapping 3D Gaussian functions.

Right now Varjo Teleport scans can only be viewed using a PC, but the company claims it can get the viewer running on standalone VR headsets too in the future.



0:00
/1:01



Scanning a scene requires you to walk around a scene for around 10 minutes while holding your iPhone up. The capture then gets processed in the background on AWS cloud servers, which takes a couple of hours. Varjo is offering this for free to those who register for early access for now, planning to monetize further in the future.

“From single rooms to town squares, captures are transformed into stunning full scale digital twins featuring accurate lighting, shading, textures, details and reflections.”

The Varjo Teleport app currently requires an iPhone Pro or iPad Pro model, but the company will soon make it available for regular iPhones and iPads, and eventually bring it to Android too.

The scans can be viewed in the Varjo Teleport PC VR app, or exported for use in Unity and Unreal. And they are measured in megabytes, not gigabytes.

A key use case could be to remotely walk around a captured scene as avatars with another person anywhere on earth, hence the name Varjo Teleport.



0:00
/0:30



A Varjo Teleport Gaussian splat scene.

I was able to try a brief demo of Varjo Teleport last week, and found the quality to be slightly but notably superior to Gracia on PC viewing captures from Luma.

However, it currently requires a very beefy PC to view in real-time at native frame rate, which makes me question Varjo’s claims it can get it running on standalone headsets. The company seems confident it can though, and we’ll keep an eye on its new Gaussian splitting effort as it evolves.

Resources