Luma Dream Machine
Luma's Dream Machine is an AI video and image platform built on the Dream Machine and newer Ray models — type a prompt or upload a photo and get short video clips, plus image generation and editing. It is aimed at creators making cinematic clips, product shots and social video without a camera or an editing timeline. (Luma now sells two different pricing ladders — see Pricing.)
Pricing varies — Luma lists two plan ladders; check lumalabs.ai/pricing before buying.
How to use Luma Dream Machine
- Open Luma and describe what you want to create in plain words: you do not need to pick a model, because the Luma Agent selects the most appropriate one for the request.
- Start from text or an image: animate a still image into video, add camera motion, extend scenes, or transform a video into a new style.
- Refine the result with the agent: generate multiple variations from one prompt, keep characters, products, and brand elements consistent with master reference assets, and edit specific images or video.
- Finish the piece on the same canvas: add voiceovers, music, sound effects, and captions, sync a character's lips to speech, and share the canvas with collaborators for review.
Frequently asked questions
Is Luma Dream Machine free?
Luma runs on a monthly subscription with a credit allowance included in each plan: Plus is $30 a month with 10,000 credits, Pro is $90 a month with 40,000 credits, and Ultra is $300 a month with 150,000 credits. Credits reset each billing cycle and unused credits expire.
What is Dream Machine now?
The Dream Machine page now presents Luma Agents, a multimodal creative workspace on an infinite canvas built around the Luma Agent. It gives access to Luma's own Ray 3.2 and Uni-1 models as well as Seedance, Kling, Veo, and MiniMax.
Can I make videos from images?
Yes. You can animate still images into video, add camera motion, create transitions between frames, extend scenes, and apply style transformations, then combine the result with audio, voiceover, and captions.
Pros & cons
Pros
- Multi-model access in one workspace: Ray 3.2, Uni-1, Seedance, Kling, Veo, and MiniMax, with the agent choosing the best model per task.
- Ray 3.2 offers frame-by-frame directing, continuity across cuts, and cinematic quality, while Uni-1 keeps brand, character, and style references consistent.
- Built for scale: parallel generation of many variations, lip sync, voiceovers, music, and collaboration on a shared canvas.
Cons
- Credit-based and pricey at the top: unused credits expire at the end of each billing cycle, and Ultra costs $300 a month.
- Like all AI systems it can make mistakes or misunderstand instructions, so outputs need reviewing and iteration — Luma's own FAQ says this.