Skip to content
Other

JOKER141/H3-Dance-LoRA

Dance choreography adapter pack for MiniMax H3

2 min readupdated

This release includes two compatible add-ons for the MiniMax H3 base video generation model, built specifically for accurate dance movements. They fix stiff, sluggish upper-body motion and lock in hand detail, letting you generate confident solo and group routines in TikTok, K-Pop, and hip-hop styles. An additional concept slider (a tunable control module) handles beat accents and rhythmic timing. The pack is a great fit for video creators experimenting with music videos and digital choreography.

Key features

  • Structured choreography: replaces the base model's erratic flailing with crisp moves from popular dance styles.
  • Clean hands: prevents fused palms and distorted fingers during fast transitions.
  • Body and clothing physics: improves torso and hip articulation along with the momentum of loose pants and long hair.
  • Tight beat sync: a dedicated slider sharpens hard stops on downbeats and boosts overall hit dynamics.

What you can do

  • Generate a solo dance: run the dance LoRA adapter at full strength for detailed single-character choreography.
  • Stage a group routine: dial down the adapter weight to generate synchronized multi-person dances without bodies merging.
  • Punch up the rhythm: apply the concept slider with a positive weight for sharp, accented hits on heavy bass drops.
  • Smooth out the flow: push the slider into negative values for softer, more continuous choreography.

Example

Recommended setup: pair the Dance Choreography LoRA at a weight of 0.7–0.9 with the Dance Beat-Sync Slider set to +0.7..+1.2 for expressive, snappy dance dynamics.

Read moreDetails
Base model: MiniMax/MiniMax-H3
Library: minimax-h3
Compatibility: fully compatible with other style and character LoRAs
Limitation: complex choreographed routines still require generating multiple seeds (cherry-picking) due to the base model's inherent randomness
Discuss Telegram 2 92

Will it run?

Working out the memory needs

This is a LoRA: it only runs on top of the base model MiniMax/MiniMax-H3. The base model's size is still being collected; the estimate will appear within a day.

Massed ComputeOpen Massed Compute

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # JOKER141/H3-Dance-LoRA_Beat-Sync-Concept-Slider
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download MiniMax/MiniMax-H3 --local-dir ~/models/MiniMax-H3
    hf download JOKER141/H3-Dance-LoRA_Beat-Sync-Concept-Slider --local-dir ~/models/H3-Dance-LoRA_Beat-Sync-Concept-Slider
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to MiniMax-H3
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

MiniMax H3 prompts

All prompts

Ready prompts with the resulting video: copy and try.

Model family

Requires the base model

MiniMax/MiniMax-H3

This model is a LoRA / adapter. Download the base model first, then load this one on top.

New posts about versions of MiniMax-H3 (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

JOKER141/H3-Dance-LoRA_Beat-Sync-Concept-Slider
Author
JOKER141
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

922

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments