Upload three reference images in order, then describe the scene in one sentence. Use “image 1 / image 2 / image 3” to name a specific subject. Chinese and English input are supported; include a short exact line if dialogue is wanted.

Choose an aspect ratio: 16:9, 9:16, 1:1, 4:3, 3:4, or 21:9. Choose Low (480p), Medium (576p, default), or High (768p) output quality. The three quality levels use different amounts of compute credits; higher quality generally costs more. Low is best for previews and may not preserve the original pose or facial features well; choose Medium or High when fidelity matters.

Output is approximately 5.17 seconds at 24 fps, with audio. The model is fixed to MiniMax H3 REF2V.

按顺序上传三张参考图，写一句想发生的事。需要指定角色时写“图1/图2/图3”，需要对白时直接给出简短台词。

视频比例可选16:9、9:16、1:1、4:3、3:4或21:9。画质分低档480p、中档576p（默认）、高档768p；三档算力消耗不同，画质越高通常消耗越多。低档适合快速预览，较难保留原图姿势和面部特征；重视还原请选择中档或高档。

输出约5.17秒，24fps，带声音。模型固定为MiniMax H3 REF2V。