ComfyUI node pack for HyperFlow [Video Rebirth's] 8-step LoRA for MiniMax-H3 (video + audio), ported onto ComfyUI's native MiniMax-H3 model. No ComfyUI core files are modified; no custom sampler is needed.
MiniMax H3 多模态 LoRA 训练、审计与交付技能：Ref2VA、联合音视频、低显存后端与十步证据链。
MiniMax H3 全域提示词工程 Skill v1.7｜AIMixer Director 导演台生产规则｜T2VA / I2VA / FL2VA / L2VA / Ref2VA｜Subject 与 Picture 语义分离、跨镜头因果桥、人物一致性、视频编辑、动作参考、音频控制、Prompt 修复｜ComfyUI / RunningHub / Codex
To associate your repository with the ref2va topic, visit your repo's landing page and select "manage topics."