Hailuo 新模型 MiniMaxH3 实现跨风格精准口型同步
Hailuo 的 MiniMaxH3 把真人和动画角色放在同一对话里练口型，表情手势都自然，和普通视频生成不同。
Hailuo 新模型 MiniMaxH3 主打精准口型同步，支持所有角色和画风。测试者 Stéphane 生成一段真人与动画角色的对话序列来评估模型，关注对话动态、表情和自然手势。
视频显示 MiniMaxH3 在跨画风对话场景中口型匹配流畅，肢体动作自然。
Precision lip-sync for every character, across all styles👄 Stéphane @steftranquillin A good model testing is to generate a dialogue sequence and see how the model performs when it comes to dialogue dynamics, expressions, natural gesture, etc. Here's how @Hailuo_AI new model #MiniMaxH3 deals with a dialogue sequence between a real person and an animated one.
This report is categorized under AITOP's "Models" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include MiniMaxH3, Hailuo, lip sync, video generation.
Source attribution: curated from Hailuo AI (@Hailuo_AI). Cross-source corroboration: 7 independent sources have covered this development (including lmarena.ai, marktechpost, orange.ai), confirming industry-wide signal strength.
