moonvalley/marey models

marey by Moonvalley — AI Model Family

marey is Moonvalley's flagship generative video model, purpose-built for professional filmmakers and creative studios seeking cinematic-quality video generation with studio-grade control. The model family addresses a critical need in modern content production: enabling creators to generate visually consistent, high-fidelity videos from text or image inputs without compromising on creative direction or ethical data sourcing.

Moonvalley, founded in 2023 and based in Toronto, Canada, developed marey specifically for the film and video production industry. Unlike many generative video tools trained on scraped internet data, marey is trained exclusively on licensed, high-resolution footage—eliminating copyright concerns and ensuring commercially safe outputs. This ethical foundation makes marey particularly valuable for studios, agencies, and independent filmmakers who need to protect their intellectual property and client interests.

The marey family currently includes two primary model variants: Text to Video and Image to Video, each optimized for different creative workflows.

marey Capabilities and Use Cases

Text to Video enables creators to generate complete video sequences from detailed text prompts. This model excels at translating narrative descriptions into visual storytelling. For example, a filmmaker might input: "A cinematic wide shot of a misty forest at dawn, with golden sunlight filtering through tall pine trees, soft camera pan to the right, birds chirping in the background"—and marey produces a coherent, atmospheric 10-second sequence at 1080p resolution.

Image to Video transforms static images into dynamic video content, preserving visual consistency while adding motion and depth. This workflow is particularly powerful for shot-by-shot filmmaking. A director can upload a carefully composed still frame—perhaps a character in a specific pose or a meticulously designed set—and marey animates it with precise motion control. This approach ensures locked-in backgrounds, consistent character appearance, and pixel-perfect art direction across multiple shots.

Both models support up to 10-second generations at 1080p resolution, with marey's advanced motion generation and physics simulation creating videos with rich atmosphere and natural movement. The models can be chained together in production pipelines: start with text-to-video for establishing shots, then refine with image-to-video for close-ups and character moments, building cohesive scenes shot by shot.

Technical specifications include support for detailed prompt adherence, iterative refinement capabilities, and attention to temporal consistency—critical for professional video work where flickering, discontinuity, or character inconsistency can derail a production.

What Makes marey Stand Out

marey distinguishes itself through three core strengths:

Ethical Data Foundation: Trained exclusively on licensed footage, marey eliminates the legal and ethical risks associated with scraped-data models. This makes it ideal for commercial productions, agency work, and any context where content provenance matters.

Studio-Grade Control: marey transforms 2D scenes into 3D environments with precise control over camera movements, object trajectories, and physics simulation. Filmmakers can specify exact motion paths and environmental behavior, not just hope the model interprets their intent correctly.

Cinematic Quality: The model is optimized for professional storytelling, generating videos with rich motion, atmospheric depth, and visual continuity. It's designed for creators who view AI as a production tool, not a novelty—those building actual films, commercials, and branded content.

marey is ideal for independent filmmakers seeking faster pre-visualization, studios automating shot generation, creative agencies producing client work at scale, and any professional who needs commercially safe, legally defensible video generation.

Access marey Models via each::labs API

The entire marey model family is accessible through each::labs, the unified AI model platform designed for developers and creative technologists. Rather than juggling multiple APIs and authentication systems, you access both Text to Video and Image to Video variants through a single, consistent interface.

each::labs provides three ways to work with marey:

• API: Integrate marey directly into your applications and production pipelines with straightforward REST endpoints
• Playground: Experiment with both model variants interactively before committing to production workflows
• SDK: Build sophisticated video generation workflows in Python or JavaScript with full control over parameters and batch processing

Sign up to explore the full marey model family on each::labs and unlock professional-grade video generation for your creative projects.