Vidu multi-shot workflow: when to use Reference to Video

To make a multi-shot film with Vidu, first ask whether a character or product repeats. Image to Video is simpler for animating one product image; Reference to Video suits repeated subjects. This is a proposed process based on official documentation, not a hands-on test.

Vidu distinguishes animating a single image or keyframes from preserving a subject across shots. Record the shot ID, subject, action, framing, intended length and transition. A one-shot product animation rarely needs elaborate references. For recurring characters, collect consistent images with permission to use them.

First, define acceptance criteria per shot. Second, choose the mode. Vidu's website describes uploading 1–7 reference images, adding motion instructions, adjusting settings, previewing and downloading. The Q4-preview API specifies 1–15 images: limits differ by interface and model. Third, generate the hardest shot first; inspect faces, clothing and logos. Consistency claims are not guarantees. Fourth, assemble clips in an editor with captions, brand text, licensed music and transitions, checking text and pacing. Fifth, fix errors by type: align reference images for subject drift, simplify motion prompts for faulty actions, or split shots differently for poor transitions. Track shot ID, reference version, first defect, retries, pass/fail and editing time.

Vidu distinguishes Image to Video, Reference to Video and Text to Video. The Vidu 2.0 Image to Video API uses one starting image and supports 4 or 8 seconds, while Q4-preview has other image limits. Verify formats, duration, resolution, credits and task statuses. Add retries and human quality checks.

As of October 30, 2026, Vidu's Chinese pricing page lists Standard at US$99 billed annually, 800 credits per month and commercial rights for paid plans. A displayed monthly equivalent is not monthly billing; generation counts vary by model and settings. Make a three-shot sample featuring one subject and measure attempts, rework and editing time. Projects needing frame-perfect control, allowing no subject changes or without editing capacity cannot rely on Vidu as an end-to-end production system.