Wan 3.0 Prime R2V Uncensored (Video)

## Overview Wan 3.0 Prime R2V Uncensored is a reference-to-video generator offered by Wiro-Partners. It turns a written scene brief plus visual and audio references into a single MP4 clip. It supports longer shots than earlier Wan versions, up to 30 seconds per run. It’s useful when you need character or style consistency across frames, not just a one-off motion clip. ## What you can build - Character-consistent short films from a small set of reference images - Product hero shots that keep logos, materials, and shape stable through motion - Multi-shot story beats guided by mixed references (images + short video snippets) - Style-locked social clips where the “look” stays consistent across variations - Prompt-driven motion tests for storyboarding and previz - Prompt + audio guided clips where rhythm and timing follow a reference track ## Inputs - A required text prompt that describes the scene and motion. You can refer to uploaded references by index, like “Image 1” or “Video 1.” - Optional reference images to lock identity, wardrobe, props, or style. You can submit up to 10 images. - Optional reference videos for motion, vibe, or shot pacing. You can submit up to 5 videos, with 15 seconds total reference duration. - Optional reference audio clips to guide voice, ambience, or timing. You can submit up to 5 audio clips, with 15 seconds total reference duration. - A target duration in seconds. You can choose any whole number from 2 to 30 seconds. - A resolution tier. Choose 480p, 720p, or 1080p. - An aspect ratio selection. “Adaptive” lets the model infer framing from the prompt and references. You can also force common ratios like 16:9 or 9:16. - An optional toggle to include an audio track in the output. When disabled, the output can be silent. - An optional toggle that rewrites and expands short prompts to add detail. - An optional numeric seed to reproduce results. Use an integer from 0 to 2,147,483,647. ## Outputs - One MP4 video clip at the chosen duration and resolution. - 30 FPS playback for Wan 3.0 outputs. - Framing that matches the chosen aspect ratio, or “adaptive” framing when enabled. - Optional embedded audio when audio is enabled. Audio is generated on the same timeline as the visuals. ## Limitations - Output length is limited to 30 seconds per run. - Reference inputs are capped. Use up to 10 images, 5 videos, and 5 audio clips. - Reference video and reference audio have a combined “short clip” budget. Keep each set within 15 seconds total. - On-screen text and fine audio texture can still degrade in complex scenes. Plan for human review. - Vague prompts can drift. Wan-style prompting often needs explicit subject, action, camera, and constraints. - Low-quality or messy inputs can break motion. Blurry images, heavy artifacts, collages, and text overlays can cause glitches. - This is a hosted model family. There are no official downloadable Wan 3.0 weights for local inference. ## Safety & compliance - “Uncensored” does not mean unrestricted use. You must follow applicable law and platform rules. - Adult content must be legal and consensual. Any CSAM or illegal content is forbidden. - Only upload references you have rights to use, including images, video, and audio of real people.