Luma Models

Luma Labs is building a next‑generation creative platform that blends computer vision, generative AI, and 3D reconstruction to let creators capture, imagine, and generate real-world scenes in richly detailed formats. Their model ecosystem includes powerful NeRF-based reconstruction, image/video generative models, and a cloud API that makes 3D capture accessible for individuals and businesses.

At the core of Luma’s technology lies Neural Radiance Fields (NeRF) — a technique that reconstructs a 3D volumetric scene from 2D input, such as video walkthroughs or multiple photographs. With Luma’s NeRF model, you can scan objects or entire scenes using your smartphone, then export usable 3D assets (like GLB or OBJ) to game engines, AR/VR experiences, or 3D design tools.

On the generative side, Luma offers Dream Machine, a text-to-video model that turns text prompts (or still images) into visual scenes with motion, lighting, and cinematic composition.

This is part of Luma’s vision to build “world models” — AI that can not only see and understand the world but create it dynamically in video form.

Luma also provides a developer-focused API, giving access to their NeRF and meshing models at scale. This means companies can integrate high-fidelity 3D reconstruction into apps, websites, or pipelines: from e-commerce (3D product previews) to virtual production and VFX.

Overall, Luma Models are designed for creators, developers, and businesses who want to move beyond flat images. Whether you’re scanning real-world objects to build a 3D scene, or generating video from imagination, Luma’s AI makes it faster, cheaper, and more accessible than traditional 3D workflows.