WAN 2.7 Uncensored Video Generator

Overview

WAN 2.7 Uncensored by Wiro Partners is an image-to-video model based on Alibaba’s Wan 2.7 generation. You provide a still image as the first frame plus a motion prompt. The model synthesizes new frames that follow your action and camera direction. It outputs a ready-to-use MP4 clip, so a single image becomes a short video.

What you can build

• Animate portraits into slow turns, facial expressions, and subtle pose changes
• Turn product photos into short ad-style clips with camera push-ins and light sweeps
• Create TikTok or Reels loops from a single key image
• Generate short fashion, fitness, or lifestyle shots from one still frame
• Make “before/after” style motion beats by re-running with different seeds
• Produce multiple takes of the same shot by changing the motion prompt only

Inputs

• A starting image file that becomes the first frame. Use JPEG/JPG, PNG without transparency, BMP, or WEBP. Keep it 240–8000 px, within a 1:8 to 8:1 aspect ratio, and under 20 MB.
• A required motion prompt in plain text. Describe what changes over time, not just how it looks.
• A required clip length choice: 5, 10, or 15 seconds.
• A required output quality choice: 720p or 1080p.
• An optional negative prompt in plain text. Use it to block specific artifacts or unwanted elements.
• An optional audio track upload in WAV or MP3 format. Keep it 2–30 seconds and under 15 MB.
• An optional audio toggle. Turn it off to force a silent video. Some deployments may ignore uploaded audio when auto-audio is enabled.
• An optional prompt rewrite toggle. Turn it on when your prompt is short or underspecified.
• An optional seed number for repeatable outputs. Use an integer from 0 to 2,147,483,647.

Outputs

• One MP4 video clip.
• The clip matches your input image framing and aspect ratio.
• The duration matches your selected length (5, 10, or 15 seconds).
• The frame size matches your selected tier (720p or 1080p).
• Audio is included only when audio is enabled. Otherwise the clip is silent.

Recommended settings

• Pick 5 seconds when you need the cleanest motion and least drift.
• Use 720p for iteration and prompt testing. Switch to 1080p for final exports.
• Turn on prompt rewrite when your prompt is under 1–2 sentences.
• Keep negative prompts short. List 5–20 targeted defects, not long blocklists.
• Set a fixed seed when you need repeatable framing and character details.

Limitations

• Output length is capped at 15 seconds per generation.
• Fine details can shimmer across frames, especially at 1080p.
• Text inside the scene is often unstable across frames.
• Extreme motion requests can cause warping, duplicated limbs, or identity drift.
• Low-quality inputs increase artifacts. Blurry photos, heavy JPEG noise, and busy collages often break motion.
• Inputs with transparency are not supported for PNG. Flatten the image first.
• Very wide or very tall images must stay within the 1:8 to 8:1 ratio limit.

Safety & compliance

• “Uncensored” usually means fewer prompt refusals, not unlimited permission.
• Don’t generate illegal content. Don’t generate sexual content involving minors.
• Don’t create non-consensual sexual content or exploitative content.
• Don’t impersonate real people for deception. Get explicit permission for likeness use.
• Only upload images and audio you own or have rights to use.