Welcome to Luma Agents
In Luma, all your work is organized on boards where you can generate, iterate, and leverage creative assets with an agent that orchestrates all the top generative image and video models for you so you never have to spend time writing a page-long Seedance 2.5 prompt.
Each board is an infinite canvas packed with helpful tools that you can access through a rich UI for granular control, or in your agent chat describe what you want and let your agent take control of production and prompt engineering
The Three Core Layers
1. Project Organization
Boards organize creative initiatives visually so teams can structure scenes, concepts, references, and production assets in one place
- Assets, versions, and explorations can be grouped and compared while preserving project context
- Teams collaborate in real time with agents advancing work alongside them
- Semantic search instantly locates assets, references, and iterations across the board
- Work evolves spatially across boards rather than through folders or disconnected timelines
2. Creative Agents
Luma's Creative agents execute and advance professional creative work across the entire process while maintaining shared project context in every board.
You:
- Direct creative work through natural language
- Request iterations, refinements, and alternative directions
- Gather references, research, and inspiration within the project
- Evaluate outputs and guide quality improvements
Agents:
- Route tasks across industry-leading image, video and audio models automatically or selectively
- Generate text, image, video, and audio assets
- Iterate and evolve assets while maintaining contextual memory across the project board
- Evaluate outputs and advance work toward final delivery.
3. Capabilities
Creative agents access and orchestrate industry-leading GenAI models to execute creative work across all modalities while maintaining shared project context. Agents route tasks across advanced video model families including Ray, Veo, Kling, and specialized image models including Uni-1, Nano Banana, Seedream, and GPT Image
Image
- Photorealistic generation
- Stylized illustration
- Concept art and visual exploration
- High-fidelity editing and enhancement
- Outpainting and reframing
Video
- Text-to-video generation
- Image-to-video animation
- Video transformation and modification
- Keyframe interpolation
- Video extension and scene continuation
- Character-consistent generation
Audio
- Sound effects generation
- Music generation
- Voiceovers with emotional control
- Word-level transcription
Lip Sync
- Audio-to-mouth synchronization
- Expression control
- Character animation
Composition
- Programmatic editing of video and motion elements
- Captions, transitions, and compositing
Agents automatically route tasks across the appropriate models while preserving context across the project. Advanced users can also access individual models through the board UI when manual control is preferred