Microsoft unveils AI tool to turn 2D images into 3D models: How to use

Available worldwide via Copilot Labs, which is a space for experimental features, Copilot 3D enables quick conversion of static images into 3D models

Microsoft has unveiled an artificial intelligence-powered tool to transform 2D images into real-life-looking 3D models. The US technology giant has dubbed it Copilot 3D and said that it is designed to make 3D creations fast, accessible, and intuitive. Copilot 3D is part of Copilot Labs and is currently available to users globally who have signed in with their personal Microsoft Account. The company recommends using it on a PC. However, it is also accessible via mobile browsers. Notably, other companies like Meta, Apple, and Nvidia are also in the race to create such a tool.

What is Copilot 3D

Copilot 3D, available only through Copilot Labs, is an AI-driven tool that converts a single image into a complete 3D model – no expertise required. Built to simplify and speed up the 3D design process, it aims to make creation more accessible and intuitive. From exploring creative concepts to prototyping ideas or aiding interactive learning, Copilot 3D encourages experimentation without the steep learning curve of conventional 3D programs.

 

For the ones who are worried about privacy, Microsoft has clarified: “Uploaded images are used by Copilot to generate your 3D models and process your request. At this time, Microsoft does not use uploaded images for training or personalisation.”

 

Notably, at this time, Copilot 3D supports only 2D image-to-3D image generation. This essentially means that it cannot create a 3D image just on the basis of a text prompt.

How to use Copilot 3D

• Open the Copilot 3D website on your PC’s browser
• Register with a Microsoft account
• Upload an image within 10 MB, and click on Generate to create the 3D model of the image.
• Download the 3D model in GLB format.

Any other players in the market?

Apple

Earlier in May, Apple previewed an AI model – Matrix3D – that can build 3D scenes using images. Furthermore, with the arrival of iOS 26 public beta, the Photos app will let users convert any regular image into a spatial image, even if the photo was originally taken with a different smartphone. 

Meta

It had published a research paper detailing “Meta 3D Gen AI” system, which it said is capable of generating 3D models using a text prompt. A research paper from Meta outlines Meta 3D Gen, a dual-AI system combining Meta 3D AssetGen and Meta 3D TextureGen. It can generate 3D characters, props, or scenes from text prompts, or apply textures to an existing 3D mesh provided by the user.

Nvidia

Nvidia’s neural radiance fields, or NeRF, can also create 3D images based on 2D images. Nvidia, in a blog earlier, said that NeRF uses neural networks to represent and render realistic 3D scenes based on an input collection of 2D images.

 

Several other companies are also making such tools that are capable of turning a 2D image into its 3D version.