Microsoft tests turning pictures into 3D models in Copilot

Microsoft’s Copilot 3D experiment continues to progress, now enabling users to duplicate pre-generated 3D objects via a new “Recreate” button. Selecting this leads to a “My Creations” section, where users can view their generated objects, browse them, and interact with the files.

The experiment also introduces the option to upload an image and produce a corresponding 3D object; however, initial testing suggests the fidelity of these generated models is still lacking, pointing to ongoing development on Microsoft’s side to address quality concerns. Users interested in quick 3D prototyping or digital artists might find these early features promising, though practical utility will depend on how the model generation improves.

Copilot 3D’s main appeal lies in offering AI-powered 3D asset creation directly from Copilot’s interface, likely under an experimental “labs” section. This hands-on generation and download flow sets it apart from most current AI tools, as Microsoft targets users ranging from hobbyists to those working in gaming, design, or education—anyone who needs simple, fast 3D content.

Despite this progress, there’s no clear indication of when (or if) the feature will move beyond private testing. Microsoft’s track record with Copilot Labs projects shows that some, like Copilot Characters, have been deprioritised or transformed; characters, for example, seem limited to voice features and haven’t seen visible updates in months. The fate of Copilot 3D will likely hinge on user interest and the technical ability to generate useful, high-quality assets, as well as Microsoft’s willingness to maintain experimental features in production.