What does “reference-driven consistency” mean?

Reference-driven consistency means faces, outfits, products, and environments drift less between shots when you give Wan 2.7 clearer anchors — reference images, clips, or start/end frames — instead of relying on one vague prompt. The model uses those anchors to lock identity, style, and composition across the full sequence, so multi-shot videos feel like the same world and the same characters from beginning to end.