Moonvalley Marey - Text to Video

Moonvalley Marey text-to-video in one workflow.

Write a detailed scene description, pick your dimensions and duration, hit Run. Marey returns a cinematic 1080p video at up to 10 seconds. Because the model was trained only on licensed footage, the output is commercially safe by default.

No image input needed. Describe the shot, set a few options, get a finished video back.

How do you prompt Moonvalley Marey for text to video?

Marey responds best to prompts written like shot descriptions. Use a five-part structure: camera movement, scale or perspective, core visual, environmental details, and lighting or technical specs. Aim for 50 words or more. Short object descriptions consistently underperform detailed cinematic prompts on this model.

Prompt Lead with camera work and angle. "Tracking shot from the side," "low-angle dolly in," "aerial sweep." Then describe the subject and scale, then the environment, then the lighting or lens. Include film-language specifics like "35mm," "shallow depth of field," "soft overcast light," or "handheld motion." The more directional your prompt, the more directional the output.

Dimensions Default is 1920x1080. Want a widescreen cinematic frame? Stick with 1920x1080 or 2048x1152. Shooting for social or portrait? Pick a 9:16 vertical option. The aspect ratio shapes the composition the model generates, so match it to where the video will end up.

Duration Choose 5s or 10s. Start with 5s when you're dialing in a shot. Once the prompt is landing the composition you want, bump to 10s for the final render. 10s costs more credits but gives the motion more time to breathe.

Negative prompt The default list already filters out the common failure modes: synthetic look, scene cuts, low-poly shading, bad rigging, AI artifacts, bloom spam. Leave it alone unless you're seeing a specific problem. If faces look waxy, add "plastic skin, rubbery face." If motion stutters, add "stiff animation, frame drops."

Seed Set to randomize for exploration. Once you hit a composition you like, pin the seed and iterate on the prompt instead. This is the fastest way to A/B test small wording changes without losing the frame you liked.

Guidance scale Default 3.5. Lower values (2 to 3) let Marey interpret your prompt more loosely, which helps when your description is short or abstract. Higher values (4 to 5) pull the output closer to the literal prompt, which helps when you need specific elements to appear. The catch: push it too high and motion gets stiff.

What is Moonvalley Marey good for?

Marey is built for professional film and ad production. It's the first video model trained only on licensed footage, so output is commercially safe for client work. Strengths: cinematic camera motion, realistic lighting, coherent physics, and strong prompt adherence at native 1080p. Best for pre-viz, B-roll, ad spots, and short narrative shots.

Commercial shoots where IP safety matters. Marey's licensed-data training removes the copyright question that blocks other video models from client deliverables. Ad agencies and studios can ship the output in paid work without a legal review holding it up.

Cinematic shots that need directorial intent. Camera moves, scale, lighting, and motion physics all hold up under scrutiny. Tracking shots look like tracking shots. Handheld looks like handheld. This is the thing Marey does better than the open-source alternatives.

When to reach for something else: fast iteration at low cost (Marey bills per generation), stylized or anime output (the model leans photorealistic), long-form sequences (5 and 10 seconds is the cap per clip). For those, a local video model like Wan or LTX will serve you better.

FAQ

What makes Moonvalley Marey different from Sora, Veo, or Runway? Marey is trained exclusively on licensed footage, which means output is commercially safe for paid client work without copyright exposure. It's also built around filmmaking controls like camera motion and scale rather than pure text-to-video. Tradeoff: clips cap at 10 seconds and cost more per generation.

What resolution and frame rate does Moonvalley Marey output? Native output is 1080p at 24fps, the standard for film and most professional video. This workflow re-encodes frames at 30fps during save, which makes the file play smoothly on web platforms and social apps without dropping frames.

How long can a Moonvalley Marey text-to-video clip be? 5 seconds or 10 seconds per generation. For longer sequences, generate multiple clips with consistent prompt language and edit them together in post. Pin the seed across generations to keep the look consistent from shot to shot.

What guidance scale should I use for Moonvalley Marey? Default is 3.5 and it works for most prompts. Drop to 2 to 3 when you want Marey to interpret loosely or your prompt is short. Push to 4 to 5 when specific elements need to appear exactly as described. Higher values can stiffen motion, so test before committing.

How to run Moonvalley Marey online? You can run Moonvalley Marey online through Floyo. No installation, no setup. Open the workflow in your browser, write your prompt, hit Run. Free to try.