The lifelike and dynamic 3D Personas that have been a standout differentiator for Apple since the release of the Vision Pro have long been speculated to use Gaussian Splatting.
Now, according to CNET, they’ve confirmed in an interview with Apple that the company is indeed using the radiance field representation to power immersive avatars within the device.
While it’s not a huge surprise that Gaussian Splatting is behind it, what is surprising is the blunt disclosure. Apple rarely reveals underlying technologies, making this a notable exception. It’s a strong endorsement of both the technology’s capability and Apple’s belief in what Gaussian Splatting can achieve. Recent updates to iOS 26 have also introduced greater fidelity in reconstructions, now capable of representing more inorganic and view dependent details, such as jewelry.
It also turns out that it’s not just Personas leveraging Gaussian Splatting, but also Apple’s Spatial images. When pressed about a potential Hyperscape competitor, Apple confirmed that they already apply Gaussian Splatting to the spatial 3D conversions of photos.
As the CNET article articulates, it’s difficult to convey just how compelling the technology is on a 2D screen. When you’re in front of someone in a virtual, lifelike capacity, it creates a true sense of presence. Another use case Apple mentions is how healthcare could be revolutionized through FaceTime calls with dynamic 3D representations, or even asynchronous patient monitoring.
There’s so much to pull and extrapolate from this. Paired with the added genlock support in the newest iPhones, it’s clear Apple knows exactly where this is headed. As I’ve continued to write, we’re hurtling toward a native 3D imaging world that’s quietly rippling across industries in an agnostic and silent way. While several companies have been building and deploying privately, expect to see more players with the scale and influence of Apple announce their own radiance field representations before the end of this year.
New to Gaussian Splatting? Start here
