Vidu Q3
Reference-based video generation. Up to 7 subject references for multi-character consistency.
Vidu Q3 solves a problem most video models dodge: keeping multiple characters consistent across shots. Upload up to 7 reference images and the model maintains identity, clothing, and features across the generated video. Text-to-video and image-to-video modes.
Example outputs
“Two architects reviewing blueprints at a construction site, hard hats and high-vis vests, morning light, documentary style”
“A product showcase: three perfume bottles rotating slowly on a marble surface, studio lighting, luxury brand commercial”
“An interior designer walking through a completed living room, touching fabric samples, soft afternoon light, editorial video”
“A game cinematic: two warriors facing each other in a misty forest clearing, armour detail, slow camera orbit, fantasy epic”
“A fashion editorial: two models walking side by side down a rain-wet city street at dusk, matching outfits, cinematic slow motion”
“An automotive reveal: an electric SUV driving through a desert landscape, drone tracking shot, golden hour, commercial quality”
How it works
Describe your scene
Type a detailed prompt describing the video you want, or upload a reference image as a starting frame.
Choose your settings
Pick your resolution and duration. See the credit cost before you generate.
Generate your video
Your video is ready in 1-3 minutes. Download, iterate, or extend the sequence.
Ready to create with Vidu Q3?
Jump into the Studio and start generating. Plans from £10/month.
Choose a PlanMulti-character consistency in AI video.
Most AI video models handle a single subject well, but fall apart when you need two or more characters to look consistent. Vidu Q3 is built around reference-based generation: upload images of your characters, products, or locations, and the model keeps them visually consistent throughout the generated video. Up to 7 reference subjects per generation.
This makes it practical for workflows that other models cannot handle. A product launch video with multiple products maintaining their exact design. A storyboard animatic with two recurring characters. An architectural walkthrough where both the building exterior and interior furniture stay consistent. A brand campaign with a team of models who all need to look like themselves across multiple scenes.
Vidu Q3 supports both text-to-video and image-to-video modes. It sits in the mid-range alongside Kling O3 and Hailuo 2.3 Pro. The trade-off is clear: you pay slightly more for multi-reference consistency that no other model in the roster can match.
Up to 7 subject references
Upload reference images for characters, products, vehicles, buildings, or any subject that needs to stay consistent. The model identifies each reference and maintains its visual identity across the video. Use it for multi-character scenes, product portfolios, or any narrative with recurring elements.
Text and image to video
Start from a text prompt for full creative control, or provide a start frame image for precise composition. Both modes support the full reference system. Combine a start frame with character references for maximum control over both scene composition and subject consistency.
Questions about Vidu Q3.
Why Stensyl?
A small indie studio building creative tools the way they should be built. No VC theatre, no funnel games, no faceless support.