Sam3D

Creating 3D assets from photos traditionally takes specialized software, skilled artists, and hours of manual work. sam3d changes that. Developed by Meta, this single-image 3D reconstruction model fuses SAM 3 open-vocabulary segmentation with geometry, texture, and layout prediction to turn ordinary photos into ready-to-use 3D assets. As a research-grade model extending the Segment Anything family, sam3d can reconstruct 3D geometry, texture, and spatial layout from just one image — and supports text or visual prompts to isolate and convert specific objects into 3D assets quickly. It's a glimpse into the future of 3D content creation.