Microsoft Unveils Copilot 3D: Transforming 2D Images into 3D Models with AI

You Can Now Turn 2D Images into 3D Models With This Copilot Feature

Users need to sign in to their personal Microsoft account to access it Microsoft has silently released a new artificial intelligence (AI) feature for Copilot that will allow users to convert a static 2D image into a 3D model. Dubbed Copilot 3D, it is a new experimental feature available via Copilot Labs. The Redmond-based tech giant said that the feature will be available to all users globally, provided they are signed into their personal Microsoft accounts. While the tech giant is not the first to offer such a feature, it claims that the 3D assets generated using Copilot are print and render-ready. While the company did not make any announcement for the new feature, it has published a new post to detail it. Microsoft describes Copilot 3D as an AI-powered tool that lets users turn an image into a fully rendered 3D model. This is an experimental feature available through Copilot Labs, and no paid subscription is required. The company says the AI tool makes the process of 3D asset creation easy and intuitive. Users do not need any knowledge about 3D modelling. They can simply upload an image of up to 10MB in size, and Copilot 3D will turn it into a print and render-ready model. These outputs can also be directly integrated into their gaming, animation, 3D printing, augmented reality (AR) and virtual reality (VR) apps, and metaverse-focused digital content projects. In case a user does not know how to get started, they can also access a Microsoft-curated library featuring unique scenes to act as inspiration for users. Notably, Copilot 3D currently only supports 2D image-to-3D model generation. It does not support text-to-3D at this moment. The tech giant recommends that users access this feature via Copilot on a desktop, although it can also be accessed via mobile browsers. While Microsoft's offering comes as an accessible and user-friendly method to generate 3D models from 2D images, the tech giant is not the first to offer such capability. Last year, Stability AI released an open-source AI model, dubbed Stable Video 3D (SV3D), which performed the same function. Tencent and Meta also released their 3D asset generation models in 2024. This year, Apple researchers released an open-source model dubbed Matrix3D that can turn images into 3D models.

Microsoft launches Copilot 3D to make 3D creation faster and easier, here's how

All models are stored for 28 days and can be used in Augmented Reality apps. Soon after the GPT-5 backed Smart Mode in Copilot, Microsoft announced its new tool that can convert a regular 2D image into a fully rendered 3D model. Named as the Copilot 3D, the new tool is claimed to be designed to make 3D creations faster, accessible and more intuitive. With this, the tech giant aims to simplify the complex process of making the 3D assets. The tool, which is part of Copilot Labs, is currently available for free but to a limited number of users. If you want to access it, you will have to sign in with a Microsoft or Google account and Microsoft recommends using the desktop browser for enhanced performance. In order to get your hands on the newly introduced Copilot 3D tool, the users will have to visit Copilot.com and then open the sidebar menu. Now select Labs and click Try now under Copilot 3D. It must be noted that the tool supports PNG and JPG images under 10MB, with plans to expand file format and size support in the future. Soon after you upload an image, Copilot 3D will generate a 3D version in seconds, exporting models in the GLB format, compatible with most 3D viewers, tools, and game engines. Models are stored for 28 days and can be viewed in the "My Creations" section. The company also advises the users to use images with a clear separation between subject and background for best results. However, the tool may struggle with animals and certain objects, early testing shows strong performance with items like furniture, umbrellas and fruit. All creations can also be exported for Augmented Reality applications. The company has also warned the users against uploading images of people of copyrighted material without the permissions, stating that violations can also result in account bans. The tech giant also stated that these creations will not be used to train the upcoming AI models, but illegal content will automatically be blocked.