Apple's SHARP: Turn ANY Image into 3D in Seconds (Running Locally!)

Описание: Apple just released SHARP, an open-source model that converts single images into 3D Gaussian Splats in under a second—running completely locally on your Mac.

I built a custom interface where you can drag and drop images and explore the results in real-time with a Three.js viewer. Everything runs on-device, no cloud needed.

In this demo, I tested it on:
M.C. Escher's impossible architecture
Vermeer's Girl with a Pearl Earring
Van Gogh's Bedroom in Arles
Scenes from Severance

The depth estimation is surprisingly good, especially considering it processes each image in ~5 seconds on my M1 Max MacBook Pro.

🔗 LINKS
SHARP GitHub: https://github.com/apple/ml-sharp
SHARP Paper: https://arxiv.org/abs/2512.10685

⚙️ TECH STACK
Apple SHARP (3D Gaussian Splatting)
Python + Flask backend
Three.js for 3D visualization
Built with Claude (Anthropic)

💭 THOUGHTS
You can use this in a lot of use cases, like VR content creation, AI image generation references, architectural visualization, historical preservation...

What would you use this for?

📧 CONTACT
LinkedIn:   / robdewinter  
Instagram:   / robdewinter1  

#SHARP #Apple #MachineLearning #3D #GaussianSplatting #ComputerVision #OpenSource


---

Built on: M1 Max MacBook Pro
Processing time: ~5 seconds per image
Model: Apple SHARP (open source)
No cloud or GPU farm required!