Vidu Q2 is an advanced video generation model developed by Shengshu Technology, a Chinese AI startup with roots in Tsinghua University. Building on the company's previous multimodal research, the model is designed to deliver high-fidelity cinematic video with a focus on "AI performance." It emphasizes the rendering of nuanced human-like micro-expressions, complex emotional acting, and precise subject consistency across frames.
The model features two distinct operation modes: Pro and Turbo. The Pro mode is optimized for 1080p high-detail cinematic sequences and professional-grade visual quality, while the Turbo mode prioritizes faster generation speeds for action-packed clips and rapid iteration. Vidu Q2 supports diverse input workflows including text-to-video, image-to-video, and a multi-reference mode that can incorporate up to seven images to guide character identity, lighting, and scene layout.
Technically, Vidu Q2 incorporates advanced "camera grammar" controls, enabling realistic cinematic movements such as push-pull zooms, pans, and tilts. It typically produces videos ranging from 2 to 8 seconds in duration. The platform utilizes a unified visual engine that powers both its video and image generation capabilities, ensuring visual identity preservation when transitioning between still and moving media.