Vidu: Vidu Q2 Reference to Video API

Vidu Q2 Reference-to-Video is a reference image-driven video generation model developed by Shengshu Technology, part of the Vidu Q2 standard series. Compared to the Pro variant, this model retains the core reference consistency capabilities while offering a more cost-effective generation solution, making it suitable for cost-sensitive or high-volume production scenarios. It accepts one or multiple reference images combined with a text prompt to generate videos that maintain strong visual consistency with the provided references in terms of character appearance, style, or scene elements. Key capabilities include: Reference Consistency: Accurately extracts character appearances, stylistic traits, or scene details from reference images and maintains visual consistency throughout the generated video, effectively reducing character drift. Standard Quality Output: Built on the Q2 standard-tier architecture, it supports smooth and naturally dynamic video generation suitable for everyday content creation needs. Text-Guided Generation: Allows control over video content, character actions, and camera movements through text prompts. Multi-Reference Support: Accepts multiple reference images to aggregate features, suitable for character or scene-based video creation. Typical use cases include AI short film production, bulk digital content generation, social media content creation, and brand/IP character animation.