The Apple AI model creates photorealistic 3D views from one image in under a second and is now open‑source for developers and researchers.
Apple has introduced SHARP (Sharp Monocular View Synthesis in Less Than a Second), an AI model that converts a single 2D photo into a photorealistic 3D scene in under a second. Unlike traditional 3D reconstruction methods that require multiple images, SHARP uses just one photo to infer depth, scale, and structure.
The model represents the scene using millions of 3D Gaussians, small points of color and light, which, when rendered, create realistic parallax effects as the viewing angle changes slightly. This allows for near-instant visualization of a scene from multiple perspectives.
SHARP achieves this speed through a single neural network pass, making it far faster than conventional methods that rely on slow per-scene optimization or specialized hardware. While it enables minor viewpoint adjustments, the model does not generate unseen areas, prioritizing realistic depth over speculative content.
Potential applications include enhanced spatial photography, AR/VR experiences, and creative visual effects. Apple’s existing devices already support depth and spatial photo features, and SHARP could inform future tools or software updates.
In a significant move for the developer community, Apple has open-sourced SHARP, sharing both the model and research code. This allows developers and researchers to experiment, extend capabilities, and build new applications beyond Apple’s ecosystem.
By combining speed, visual fidelity, and accessibility, SHARP marks a major step forward in AI-driven 3D reconstruction from single images, offering both practical and experimental opportunities in technology and media.
Also Read: Mohun Bagan banned by AFC over Iran match
