Luma AI
Luma AI lets you turn real objects and spaces into 3D models using nothing but your phone. You walk around the subject while the app records it, and Luma reconstructs a detailed 3D capture you can view, share or export. The same company also makes Genie, a text-to-3D generator, and the Dream Machine video model.
Quick Facts
| Developer | Luma AI (founded by Amit Jain and Alex Yu) |
|---|---|
| First released | 2021 |
| Latest version | App and web service, updated continuously (see official website) |
| Platform | iOS app and web |
| Technology | Neural radiance fields and Gaussian splatting |
| License | Proprietary; free app with paid credit plans |
| Pricing | Free tier; paid credits for advanced generation (see official website) |
What is Luma AI?
Luma AI started as a research-driven company out of Palo Alto, founded in 2021 by Amit Jain and Alex Yu, and its first product was a phone app that captures real-world objects as 3D. The trick is a technique called a neural radiance field, or NeRF, which learns a 3D scene from a set of photos; Luma's newer pipeline also uses Gaussian splatting, which renders captures even faster.
The workflow could not be simpler: open the app, slowly walk around the object or room you want to capture, and let the cloud process the video into a 3D model. The results are photorealistic, which is why Luma became the favorite tool for digitizing everything from sneakers to full museum spaces.
The company later broadened into generative AI with Genie, a text-to-3D model, and Dream Machine, a viral text-to-video model. For game developers, the capture side remains the killer feature: real-world props and locations become game assets with almost no manual work.
Key Features
- Phone-based 3D capture of real objects and scenes
- NeRF and Gaussian splatting reconstruction for photorealism
- Web viewer for sharing and embedding captures
- Genie text-to-3D generation from prompts
- Exports to common 3D formats for engines and DCC tools
- Dream Machine text-to-video generation for related creative work
How to get started
Download the Luma app for iOS, create an account and point the camera at something with interesting texture and good lighting. Walk slowly around it, keeping the subject in frame, and Luma uploads and processes the capture. A few minutes later you have a 3D model in the app.
From the app or the web dashboard you can view, share or export the capture. For game work, export to a standard format and import into Unity, Godot, Blender or your engine of choice. For generated assets, switch to Genie and type a prompt instead.
- Install the Luma app on your iPhone
- Walk slowly around the object or scene you want to capture
- Let Luma process the capture in the cloud
- Review, share or export the 3D model
- Try Genie for text-to-3D generation from prompts
Use cases
- Digitizing real props and objects as game assets
- Capturing locations for virtual production and set dressing
- Product visualization and e-commerce 3D
- Museum and collection digitization
- VFX reference assets and matchmoving aids
- AR experiences anchored to real-world captures
Pricing and licensing
The Luma app is free to download and includes free captures, with credits consumed for processing and advanced features. Genie and Dream Machine generation also run on credits, and Luma sells packs and subscriptions when you need more.
Captures are stored in your Luma account and you can export them for your own projects. As with all cloud services, check the current terms on the official website, since pricing and credit details evolve as the product lineup grows.
Pros and cons
Luma's magic is turning a casual phone walk-around into a photorealistic 3D model. No special hardware, no scanning rig, no manual reconstruction: the quality per effort is hard to beat, and the web viewer makes sharing captures trivial.
The limitations are real too: capture quality depends heavily on lighting, texture and a steady hand, very shiny or transparent objects confuse the pipeline, and heavy generation work costs credits. And for real-time game assets, photorealistic captures often still need decimation and cleanup.
Alternatives
- Polycam: popular phone-based 3D scanning with LiDAR support
- RealityScan by Epic Games: photogrammetry for high-fidelity captures
- KIRI Engine: affordable photogrammetry app with cloud processing
- Meshroom: free, open-source photogrammetry for desktop
- Meshy and Rodin: AI generation when you want models from scratch