Character-Consistent Scene Generation With Reference Images
Wan 3.0 video prompt by Wan (Alibaba) · original post
A clip showing a male and female character in a lakeside reed field, with their appearances and the sunset scene strictly matching uploaded reference images.
the prompt
Based on the uploaded character card, strictly maintain the male character's features: black hair in a high bun, sideburns with stray hairs blown by the wind, sharp dark eyes, a thin and sharply defined facial contour; strictly maintain his costume details: gray-white cross-collar inner robe, outer black-gray tattered fringe shoulder cape, multiple silver coin pendant chains on the chest, deep brown wide belt wrapped around layered fabric. Based on the uploaded character card, strictly maintain the female character's features: deep brown long straight hair slightly damp and falling over the shoulders, cool side-profile contour, dark pupils; strictly maintain her costume details: olive-brown leather stand-collar short martial attire, multiple cross-bound leather straps on the chest and abdomen, inner light-colored shirt collar. The scene is strictly based on the uploaded lakeside reed beach image (golden reed clusters swaying en masse in backlit sunset light, ground scattered with dark gravel and rocks, lake surface shimmering with fine ripples in the distance, sky transitioning from cyan to gold forming a band of warm-cool交界 color), do not replace or mix in other scene elements.
Made this? Fix the credit or remove it.














