This might be the coolest AirDrop you've ever seen.
An XR developer just created an interaction that lets you grab a photograph from an iPhone with your fingers, open your hand, and step inside the scene using Apple Vision Pro.
And the entire experience combines three different AI technologies.
GPT-6.2 Astra.
Apple ML-SHARP.
Tripo AI.
Here's how it works.
FIRST: FROM A PHOTO TO A 3D WORLD
The creator, Bihe Xi, used Apple's ML-SHARP to reconstruct a three-dimensional Gaussian splat representation from a single photograph.
Unlike a normal panoramic image, a Gaussian splat can provide a sense of real spatial depth when the viewer moves around the scene.
SECOND: THE INTERACTION
GPT-6.2 Astra helped the developer build the spatial interaction.
The Apple Vision Pro recognizes a photograph displayed on the iPhone.
The user performs a pinch gesture.
Particles appear to peel away from the photo and gather in the user's hand.
Then, when the hand opens, the reconstructed environment unfolds around them.
THIRD: THE DETAILS
The floating ginkgo leaves were created using Tripo AI.
Those additional 3D assets make the transition feel more physical and atmospheric, turning a technical demonstration into something that feels almost magical.
WHY THIS IS INTERESTING
We've spent years transferring photos between our devices.
AirDrop sends the file.
Cloud storage synchronizes it.
Photo galleries let us browse it.
But what if the next step isn't simply viewing a photograph?
What if you can step into it?
Imagine revisiting a street from your childhood, a holiday destination, or a moment you captured years ago.
One photo becomes a spatial memory you can explore.
And AI coding tools are making experimental interactions like this increasingly accessible to independent creators.
IMPORTANT LIMITATIONS
This is a custom Apple Vision Pro prototype, not a new Apple AirDrop feature.
The photograph and its 3D Gaussian splat are prepared in advance. The pinch gesture triggers the spatial transition rather than generating a complete 3D scene instantly.
Also, ML-SHARP estimates three-dimensional structure from a single photograph. It doesn't recover the complete real-world geometry of everything outside the original view.
Still, the interaction is remarkable.
Because this feels like a glimpse of what spatial computing should become.
Not just floating windows in front of your face.
But natural interactions that make digital content feel like part of the physical world.
ORIGINAL CREATOR & DEMONSTRATION
Bihe Xi:
https://x.com/xbh_artist/status/2123045064280571987
Apple ML-SHARP:
https://apple.github.io/ml-sharp/
Tripo AI:
https://www.tripo3d.ai/
Would you want to walk inside your old photographs?
And which memory would you visit first?
#AppleVisionPro #GaussianSplatting #SpatialComputing #AICoding #MixedReality
YouTube / ABVX
