A video generation platform from Luma Labs. Feed it text, an image, or a video, and it generates stills and clips with camera movement, composition, and physically plausible motion. Released in June 2024, it is built around Luma’s own Ray model family, balancing fast draft generation with high-fidelity final renders. Work is organized into board-based projects, so ideation through finishing happens in one place. In 2026, Ray3.14 added native 1080p output, and in March the product expanded into the “Luma Agents” workspace that bundles multiple models together.
Key Features
- Text, image, and video inputs: Beyond writing a prompt, you can start from a still image and animate it, or transform an existing video into a different look. It is easy to build on footage and artwork you already have
- Camera moves and physical dynamics: You can direct zooms, pans, and dolly shots, along with how subjects move. This goes past “a picture that moves” toward shots that read as real footage
- Reference images and character consistency: Specify reference images or characters to keep the same look across multiple cuts, which matters when you are building a continuous, story-driven sequence
- Choice of Ray models: Pick between models tuned for fidelity (Ray3) and for speed and lower cost (Ray3.14). That makes it practical to iterate on rough cuts before committing to a final render
- Board-based project management: Lay out generated assets on a board to organize and compare them, so your trial-and-error history stays as a reusable asset
- API access: An API is available for developers who want to embed generation into their own services and workflows
Pricing
| Plan | Monthly (annual billing) | Credits | Highlights |
|---|---|---|---|
| Plus | $30 ($25) | 10,000 | Luma and partner models, guest collaborator editing, commercial use |
| Pro | $90 ($75) | 40,000 | Everything in Plus, plus 4x agent usage |
| Ultra | $300 ($250) | 150,000 | Everything in Pro, plus 15x agent usage |
| Team / Enterprise | Contact sales | Shared | Team management, project organization, usage analytics, SSO, dedicated support |
Pricing is as of August 2026. The official pricing page lists the paid plans above and does not list a free plan (check the official site for trial options). For the latest pricing, see the official site.
Pros & Cons
✅ Pros
- Accepts images and video as input, not just text, so you can build from existing material
- Camera direction gives you real control over how a shot is framed and moves
- Reference inputs keep characters consistent, making continuous scenes easier to produce
- Switching between fast and high-fidelity models keeps iteration costs down
- Paid plans include commercial use rights, so it fits professional production
- A public API lets you embed generation into existing workflows
⚠️ Cons
- Generation is credit-based, and long or high-resolution outputs consume credits quickly
- The official pricing page lists no free plan, so serious evaluation generally means paying first
- Simultaneous audio generation is weaker than some competitors; narration and sound effects usually need to be produced separately
- Results vary with how prompts are written, so hitting the exact shot you want takes practice
- Models turn over quickly, and feature support differs by model (faster models may drop some capabilities)
Comparison with Similar Services
| Criteria | Dream Machine | Runway | Pika | Sora |
|---|---|---|---|---|
| Provider | Luma Labs | Runway | Pika Labs | OpenAI |
| Main strength | Camera moves, consistency, fast drafts | Broad production suite with editing | Approachability and effects | Prompt understanding and natural motion |
| Inputs | Text, image, video | Text, image, video | Text, image | Text, image |
| Free usage | Not listed on the official pricing page | Free tier available | Free tier available | Mainly higher plans |
| API | Yes | Yes | Limited | Yes |
Who Is It For
- People who want to animate stills or illustrations into short clips for social media and ads
- Video creators who need to share a visual idea quickly at the storyboard or mockup stage
- Anyone building several cuts around the same character to assemble a simple narrative
- Creators who want control over framing and camera movement, not just subject matter
- Developers who want to add video generation to their own product
Summary
Dream Machine focuses on producing shots that hold up as actual footage, not just moving images. Camera direction and reference-based consistency are its core strengths, and the model lineup makes it practical to mass-produce rough cuts before finishing a final render. Audio generation lags some competitors, and the credit system means every experiment has a cost. A reasonable approach is to estimate the credits a few short cuts will need, start on Plus, and move up once your use case is clear.