Code for paper
"FlowerDance: MeanFlow for Efficient and Refined 3D Dance Generation"

[Paper] | [Project Page]

To set up the necessary environment for running this project, follow the steps below:

1. Create a new conda environment

   conda create -n Flower_env python=3.10 conda activate Flower_env

2. Install PyTorch (CUDA 12.8)

   pip install torch==2.7.1+cu128 torchvision==0.22.1+cu128 torchaudio==2.7.1+cu128 \ --index-url https://download.pytorch.org/whl/cu128 

3. Install remaining dependencies

   pip install -r requirements.txt

• Download the Preprocessed feature from Google Drive and place them into ./data/ folder.
• Download the Checkpoints for evaluation and place them into the ./runs/ folder:
  Download Link

After downloading the necessary files, ensure the directory structure follows the pattern below:

FlowerDance/ │ ├── data/ ├── dataset/ ├── model/ ├── runs/ ├── requirements.txt ├── args.py ├── EDGE.py ├── inpaint.py ├── test.py └── vis.py 

export WANDB_MODE=offline accelerate launch train.py --batch_size 128 --epochs 4000 --feature_type baseline

To evaluate the our model’s performance:

python test.py --batch_size 128

@article{yang2025flowerdance, title={FlowerDance: MeanFlow for Efficient and Refined 3D Dance Generation}, author={Kaixing Yang and Xulong Tang and Ziqiao Peng and Xiangyue Zhang and Puwei Wang and Jun He and Hongyan Liu}, journal={arXiv preprint arXiv:2511.21029}, year={2025} }