Vidu: Vidu 2.0 Reference to Video API

Vidu 2.0 Reference-to-Video is the second-generation reference image-driven video generation model developed by Shengshu Technology, part of the Vidu 2.0 series. Built upon comprehensive upgrades over the first generation, this model delivers significantly improved consistency between reference images and generated videos, smoother motion dynamics, and enhanced overall visual quality. It supports high-quality video generation from one or multiple reference images combined with text descriptions, and is widely used in AI content creation. Key capabilities include: Enhanced Reference Consistency: Compared to Vidu 1.0, the 2.0 version substantially improves retention of character appearances, clothing details, and stylistic features, effectively reducing visual drift during generation. Smoother Motion Performance: Optimized motion generation algorithms produce more natural and fluid character movements and cinematic camera motions. Higher Visual Quality: Supports higher resolution output with finer visual details and more accurate color reproduction. Fine-Grained Text Control: Enables precise control over video content, character actions, scene atmosphere, and cinematographic language through text prompts. Multi-Reference Fusion: Accepts multiple reference images as input, capable of synthesizing multiple visual elements into composite scene videos. Typical use cases include AI film and TV content production, digital human video generation, brand/IP character animation, creative advertising, and social media short video creation.