Product detail reference

Cited / reported data

Seedance 2.0 keeps characters and visual style consistent across shots, so your multi-shot stories stay coherent. How it works: upload a reference image of your character. The model analyzes facial features, clothing details, and overall visual style, then applies those constraints to every subsequent generation. This means a character's hair color, clothing pattern, and facial structure stay locked even as the scene, camera angle, or lighting changes between shots. When to use this: multi-shot ad campaigns where a model needs to appear in several scenes; product videos where the item must look identical across angles; serialized content like short dramas or episodic brand stories; any project where visual continuity across shots matters more than one-off creative variety. Tips and practical notes: set reference strength between 70–80% for natural-looking results that still allow slight scene-appropriate variation. Keep lighting descriptions consistent across prompts. For product shots, export the highest-resolution reference you have — fine details like text on packaging or stitching patterns are easier to preserve from a sharp source. If you notice drift after several generations, re-anchor by using the original reference image rather than a generated frame.

Across multi-shot work, common issues include characters drifting between shots, product details softening, small text blurring, or scenes losing a unified look. This page focuses on the consistency-related controls often used to reduce those issues and keep faces, clothing, and fine details more stable.