R2V, Explained
What Is R2V?
Reference to Video,
Explained in Plain English.
R2V means reference to video: the AI generation mode where the model renders a brand-new clip while staying conditioned on your own reference files — a face, a product, a style sheet, a movement clip. Here is the definition, how it differs from text-to-video and image-to-video, and where to try it.
Model Options
Choose the workflow that fits your scene
Start with a prompt, add references when needed, and pick the generation path that matches the output.
Wan 3.0 R2V
Wan 3.0's reference mode is the most complete way to feel what R2V does: upload faces and wardrobe stills, a motion clip, and optional voice takes, and Wan re-performs your new scene while carrying the identity, the movement, and the sync.
The reference-to-video hub
Three R2V engines run on Moosky — dialogue-first, motion-first, and framing-first. The hub lays the lanes out side by side so you can pick per shot.
Example Outputs
Clips made from references on Moosky
Real outputs from reference-conditioned generation — the same mode this page explains.
Why use it on Moosky AI
Powerful tools. Seamless experience.
Identity lives in your uploads
References, not adjectives, carry the character — the same face, product, and styling survive every take across the project.
Same cast, new scenes
References define who appears, not how the frame is built, so the character can walk into any setting, angle, or outfit you prompt next.
Movement you can borrow
On R2V engines that take video references, a short clip of the action you want steers the model's motion instead of its imagination.
Voice rides with the face
Attach an audio reference on supported engines and the performed line keeps that timbre, so the character sounds as consistent as it looks.
The prompt stays simple
With the cast locked by references, your prompt only has to describe the scene — setting, action, camera — instead of re-inventing the subject every run.
Pay per clip
R2V runs on Moosky credits, quoted per generation in the composer before anything renders. No subscription, and re-shoots at the same settings cost the same flat rate.
How It Works
Four steps from reference files to a finished clip
Collect your references
Gather one or two clear photos of the face, product, or look to carry — plus optional motion and voice files.
Upload and label them
Tell the model what each file carries — Image 1 is the face, Video 1 is the movement — so references play their roles.
Describe the new scene
Prompt the shot you want: setting, action, camera. The references hold the cast; your words direct the scene.
Render and re-shoot cheap
Pick an engine, watch the credit quote, and generate. Keepers ship; re-shoots cost the same flat rate.
What does R2V mean, in one honest answer
R2V is short for reference to video: an AI render conditioned on files you supply — a face, a product, a style sheet, a movement clip — instead of a description that re-invents the subject every take. Searchers typing ai r2v or asking what does r2v mean are asking one question: how does the model keep the cast from drifting? The answer is your uploads.
The decision ladder is simple. Text to video fits one-off clips with no continuity needs. Image to video animates a single still — the picture itself becomes the opening frame. R2V is the lane for projects: the references define who appears, the prompt defines what happens, and the same character survives take after take in new settings and angles.
On Moosky, Wan 3.0 R2V is the most complete example of the mode: upload faces, wardrobe stills, a motion clip, and optional voice takes, and the model re-performs your new scene while carrying all of it, billed per clip in credits with the price quoted before anything runs.
Credit-based
Simple, honest pricing
R2V is billed in Moosky credits (100 per $1) like every other generation — the composer quotes the exact cost from your clip length and resolution before anything runs, and nothing is charged for a take you discard before running it.
More from Moosky AI
Explore our complete suite of AI tools.
Frequently Asked Questions
What does R2V stand for in AI video generation?
R2V stands for reference to video. It names an AI generation mode where the model renders a new clip while conditioning on reference files you supply — photos of a face or product, style sheets, movement clips, sometimes voice takes. The prompt still writes the scene; the references decide who and what appears in it, and they stay locked from the first take to the last.
How is R2V different from text to video?
In text to video the description is everything: ask for the same character twice and you get two different people, because the model invents the subject fresh each run. R2V moves identity out of the prompt and into your uploads, so the cast stops drifting. Text to video is still the right tool for one-off clips with no continuity needs — R2V earns its keep the moment a project needs the same face in many shots.
When should I use R2V instead of image to video?
Image to video takes a single still and animates it — the picture itself becomes the opening shot, composition and all. Choose it when one frame is exactly right. Choose R2V when the character matters more than the framing: references define identity rather than layout, so you can put the same subject into new scenes, angles, and outfits and keep them recognizably themselves across the whole set of clips.
What should I upload as an R2V reference?
Start with one or two clear photos of the face, product, or style you want carried — front-facing, well lit, nothing cluttered behind the subject. From there you can add wardrobe or prop stills, a short clip of the movement you want the model to borrow, and, on engines that support it, audio takes for voice. Most teams get usable consistency from two or three good images; longer reference kits sharpen the edges.
Is R2V the same thing as ref2v?
Yes. Ref2v is just an alternate shorthand for reference to video, popularized by the spelling MiniMax uses for its R2V mode, and searchers mix the two freely. Both abbreviations describe the same capability: a render conditioned on your reference files. You will also see reference-to-video spelled out on model cards and mode names — every spelling points at the same idea.
What does an R2V workflow look like on Moosky?
Upload your references, label what each file carries (face, outfit, motion, voice), then describe the new shot in plain language — setting, action, camera. Pick an R2V engine, and the composer quotes the credit cost from your clip length and resolution before anything renders. Re-shoots and edits cost the same flat per-clip rate, with no subscription behind them.
Where can I try R2V?
Open the Wan 3.0 R2V composer for the reference-kit workflow with motion transfer and native audio, or start at the reference-to-video hub to compare the engines side by side before you pick. Both run on one Moosky credit balance, so switching models mid-project costs nothing but the credits the clip itself uses.
Disclosure
Moosky AI is an independent platform provider and reseller of AI model access. Product names, model names, and company names shown on this page belong to their respective owners. Moosky AI is not officially affiliated with, sponsored by, or endorsed by those providers unless expressly stated. Page content, examples, and previews may include AI-assisted and synthetic content.