Vidu Reference

Vidu Q1 is the newer generation model that supports native background music generation and provides superior fidelity when fusing multiple complex references. In contrast, the original Vidu model offers classic motion dynamics and does not support the background music toggle, making it primarily useful for legacy project continuity or fine-tuning comparisons.

Use Vidu Q1 for almost all multi-reference workflows, particularly when you need synchronized atmospheric audio or need to fuse distinct character, clothing, and background images cleanly. Reach for the original Vidu model only when you want to match the specific motion cadence of earlier generations or benchmark complex shot behavior.

You can upload up to seven reference images into the model. This multi-entity fusion allows you to combine separate images for character faces, distinct outfits, props, and custom background environments into a single cohesive scene.

The movement amplitude parameter controls the physical speed and camera dynamics across four settings: Auto, Small, Medium, and Large. For character portraits and close-up facial shots, use Small or Medium to avoid facial morphing and limb distortion; reserve Large for wide-angle landscape dynamics like ocean swells or heavy winds.

Avoid using Vidu Reference for text-only generation without reference images, as the underlying architecture relies heavily on visual anchors. Additionally, avoid hyper-kinetic action shots, complex acrobatics, or fine finger manipulations such as typing on a keyboard or playing musical instruments, as rapid limb movements can cause visual melting artifacts.