Jojocodex/minimax-h3-wushu-action-lora
text-to-video model by Jojocodex
Older snapshot: check the original source for current information.
From the publisher
Model card & documentation
Source previewRead the publisher’s intended use, setup instructions, evaluations and limitations. The original model card is the source of truth.
武打动作 LoRA，基于 MiniMax-H3（Comfy-Org/MiniMax-H3）训练，让模型生成人体武打/功夫动作（拳、腿、踢、棍法等），专注人体打斗动作优化。 个人在测试中4步和8步加速LORA偶尔会有画面糊的情况。Wushu-Action-LoRA 0.5的强度搭配0.3强度spatial-physics-lora，打斗的效果会更好。 使用了LORA（0.5强度） 没使用LORA wushuaction 是本 LoRA 的触发词，放在提示词最前面。训练数据中 FIGHT 类 caption 全部以它开头，模型已把它和"武术动作"绑定。 例： 第 2 段是灵魂 —— 训练 caption 全是"具体招式 + 身体协调 + 发力"的详细描述，写得越具体，动作越到位。 ① 动作主体 ② 招式与发力（核心） ③ 场景/镜头 ④ 画质词（可选） 单人拳脚 双人对打 兵器 爆发力/跳跃 自由发挥模板（替换招式词） → 例：wushuaction, a martial artist performing a flying side kick, explosive leg drive, full body extension, indoor studio, medium shot 训练时 neg 为空（模型未学负向），建议只加画质/动作负向： 触发词:kungfu combat [触发词], [主体/人物], [动作], [场景环境], [镜头语言], [风格质感], [画质词] 【示例 1:双人对打,经典武侠】 英文: kungfu combat, two martial artists in a dynamic kung fu fight, exchanging punches and kicks, blocking and dodging, cinematic action, outdoor arena, dust and motion blur, dramatic lighting, 4k, high quality 中文: 功夫对战,两名武术家在激烈的功夫对打中,拳脚相交,格挡闪避,电影级动作, 室外竞技场,尘土飞扬与动态模糊,戏剧性光影,4K,高质量 【示例 2:古装武侠】 英文: kungfu combat, ancient Chinese wuxia style, two warriors in traditional robes, flying kicks and sword clashes, bamboo forest, misty atmosphere, slow motion impact, epic music vibe, cinematic composition 中文: 功夫对战,中国古装武侠风格,两名身着传统长袍的武士,飞踢与刀剑交锋, 竹林,雾气缭绕,慢动作冲击,史诗氛围,电影构图 【示例 3:单人套路(武术表演)】 英文: kungfu combat, a solo…
Read the full model card ↗ · Preview checked 2026-10-29T02:15:45.267Z
Inside the original model card — Document outline
- 1 武打动作 LoRA · Wushu Action LoRA (MiniMax-H3).最新更新 wushuactionh3lorav5 · 3000 步。
- 测试视频（对比原本）
- 对比视频
- 1.1 触发词（必须放开头）
- 1.2 提示词结构模板（4 段式）
- 1.3 关键词库（从训练数据提炼）
- 1.4 提示词示例（可直接使用）
- 1.5 负面提示词建议
- 2 新增4500步的打斗动作LORA(这是收集的其他素材重新训练的)KungFuMiniMaxH3LoRAstep4500.safetensors
- 2.1 触发词
- 2.2 提示词结构模板
- 2.3 示例提示词(可直接使用)
Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.
Documentation belongs to its respective authors. Reported project/model license: apache-2.0. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.
Model overview
Jojocodex/minimax-h3-wushu-action-lora is a text-to-video model repository published by Jojocodex on Hugging Face. The source reports the minimax-h3 library.
This page summarizes Hub metadata. For intended use, training data, evaluation results and limitations, consult the original model card.
Model facts
- Task
- text-to-video
- Library
- minimax-h3
- Recent downloads (30 days)
- 4,158
- Cumulative likes
- 68
- Hugging Face trending score
- 3
- Reported safetensors parameters
- Not reported
- Architecture
- Not reported
- License
- apache-2.0
- Access
- Not gated by Hugging Face
- Created
- 2026-08-16T19:02:00.000Z
- Last modified
- 2026-08-30T00:37:44.000Z
Compatibility and lineage
Language coverage was not reported.
Reported base models: Comfy-Org/MiniMax-H3
Parameter count is not a RAM/VRAM requirement. Check precision, quantization, context length and runtime compatibility in the model card; no hardware or API-cost claim is inferred here.
Use and evaluate
Check license terms and access requirements first. Review the model card’s documented loading instructions, evaluations and safety limitations. Benchmark scores are not imported by this directory, and popularity is not an accuracy ranking.
No model weights are downloaded or executed by AltAPIs. Never enable remote model code without reviewing it.
Source and freshness
Source: Hugging Face Hub. Metadata observed 2026-10-18T06:44:04.504Z. Daily imports are snapshots, not real-time monitoring.
Popularity and source listings do not establish security, suitability, licensing rights or benchmark performance.
Related models
QuantStack/Wan2.2-TI2V-5B-GGUF
QuantStack/Wan2.2-T2V-A14B-GGUF