Luma began in 3D. Its first app let anyone capture an object or space with a phone camera and turn it into a photorealistic 3D scene using neural radiance fields and, later, Gaussian splatting. People used it to scan products, rooms, and sculptures, and share them in a browser or export them to 3D tools. It also released Genie, a text-to-3D model that creates 3D assets from a description.
The company has since shifted its attention to video and multimodal generation. Its Ray video models create clips from text or images, and tools for editing existing footage, keyframes, and reframing give creators control. These models are available through the Dream Machine app and an API. Luma has also introduced agents that plan and carry out creative work across text, images, video, and audio, and it works with studios on production workflows.
Luma offers a free tier and paid plans for the apps, with API pricing for developers. The 3D capture side has become less central as video took over, so people looking for 3D scanning may prefer Polycam, Scaniverse, or Kiri Engine. For video generation, Runway, Google’s Veo, Kling, and Pika are the main rivals. See the Luma Dream Machine entry for the video product in more detail.