HunyuanCustom

HunyuanCustom 🌅

HunyuanCustom: A Multimodal-Driven Architecture for Customized Video Generation

🔥🔥🔥 News!!

• June 13, 2025: 🚀 HunyuanCustom supports single GPU with only 8GB VRAM, many thanks to WanGP.
• June 6, 2025: 💃 We release the inference code and model weights of audio-driven and video-driven powered by OmniV2V.
• May 13, 2025: 🎉 HunyuanCustom has been integrated into ComfyUI-HunyuanVideoWrapper by Kijai.
• May 12, 2025: 🔥 HunyuanCustom is available in Cloud-Native-Build (CNB) HunyuanCustom.
• May 8, 2025: 👋 We release the inference code and model weights of HunyuanCustom. Download.

📑 Open-source Plan

• HunyuanCustom
  • Single-Subject Video Customization
    • Inference
    • Checkpoints
    • ComfyUI
  • Audio-Driven Video Customization
    • Inference
    • Checkpoints
    • ComfyUI
  • Video-Driven Video Customization
    • Inference
    • Checkpoints
    • ComfyUI
  • Multi-Subject Video Customization

Contents

• HunyuanCustom 🌅
  • 🔥🔥🔥 News!!
  • 📑 Open-source Plan
  • Contents
  • Abstract
  • HunyuanCustom Overall Architecture
  • 🎉 HunyuanCustom Key Features
    • Multimodal Video customization
    • Various Applications
  • 📈 Comparisons
  • 📜 Requirements
  • 🛠️ Dependencies and Installation
    • Installation Guide for Linux
  • 🧱 Download Pretrained Models
  • 🚀 Parallel Inference on Multiple GPUs
  • 🔑 Single-gpu Inference
    • Run with very low VRAM
  • Run a Gradio Server
  • 🔗 BibTeX
  • Acknowledgements

---

Abstract

Customized video generation aims to produce videos featuring specific subjects under flexible user-defined conditions, yet existing methods often struggle with identity consistency and limited input modalities. In this paper, we propose HunyuanCustom, a multi-modal customized video generation framework that emphasizes subject consistency while supporting image, audio, video, and text conditions. Built upon HunyuanVideo, our model first addresses the image-text conditioned generation task by introducing a text-image fusion module based on LLaVA for enhanced multi-modal understanding, along with an image ID enhancement module that leverages temporal concatenation to reinforce identity features across frames. To enable audio- and video-conditioned generation, we further propose modality-specific condition injection mechanisms: an AudioNet module that achieves hierarchical alignment via spatial cross-attention, and a video-driven injection module that integrates latent-compressed conditional video through a patchify-based feature-alignment network. Extensive experiments on single- and multi-subject scenarios demonstrate that HunyuanCustom significantly outperforms state-of-the-art open- and closed-source methods in terms of ID consistency, realism, and text-video alignment. Moreover, we validate its robustness across downstream tasks, including audio and video-driven customized video generation. Our results highlight the effectiveness of multi-modal conditioning and identity-preserving strategies in advancing controllable video generation.

HunyuanCustom Overall Architecture

We propose HunyuanCustom, a multi-modal, conditional, and controllable generation model centered on subject consistency, built upon the Hunyuan Video generation framework. It enables the generation of subject-consistent videos conditioned on text, images, audio, and video inputs.

🎉 HunyuanCustom Key Features

Multimodal Video customization

HunyuanCustom supports inputs in the form of text, images, audio, and video. Specifically, it can handle single or multiple image inputs to enable customized video generation for one or more subjects. Additionally, it can incorporate extra audio inputs to drive the subject to speak the corresponding audio. Lastly, HunyuanCustom supports video input, allowing for the replacement of specified objects in the video with subjects from a given image.

Various Applications

With the multi-modal capabilities of HunyuanCustom, numerous downstream tasks can be accomplished. For instance, by taking multiple images as input, HunyuanCustom can facilitate virtual human advertisements and virtual try-on. Additionally, with image and audio inputs, it can create singing avatars. Furthermore, by using an image and a video as inputs, HunyuanCustom supports video editing by replacing subjects in the video with those in the provided image. More applications await your exploration!

📈 Comparisons

To evaluate the performance of HunyuanCustom, we compared it with state-of-the-art video customization methods, including VACE, Skyreels, Pika, Vidu, Keling, and Hailuo. The comparison focused on face/subject consistency, video-text alignment, and overall video quality.

Models	Face-Sim	CLIP-B-T	DINO-Sim	Temp-Consis	DD
VACE-1.3B	0.204	0.308	0.569	0.967	0.53
Skyreels	0.402	0.295	0.579	0.942	0.72
Pika	0.363	0.305	0.485	0.928	0.89
Vidu2.0	0.424	0.300	0.537	0.961	0.43
Keling1.6	0.505	0.285	0.580	0.914	0.78
Hailuo	0.526	0.314	0.433	0.937	0.94
HunyuanCustom (Ours)	0.627	0.306	0.593	0.958	0.71

📜 Requirements

The following table shows the requirements for running HunyuanCustom model (batch size = 1) to generate videos:

Model	Setting(height/width/frame)	GPU Peak Memory
HunyuanCustom	720px1280px129f	80GB
HunyuanCustom	512px896px129f	60GB

• An NVIDIA GPU with CUDA support is required.
  • The model is tested on a machine with 8GPUs.
  • Minimum: The minimum GPU memory required is 24GB for 720px1280px129f but very slow.
  • Recommended: We recommend using a GPU with 80GB of memory for better generation quality.
• Tested operating system: Linux

🛠️ Dependencies and Installation

Begin by cloning the repository:

git clone https://github.com/Tencent/HunyuanCustom.git
cd HunyuanCustom

Installation Guide for Linux

We recommend CUDA versions 12.4 or 11.8 for the manual installation.

Conda's installation instructions are available here.

# 1. Create conda environment
conda create -n HunyuanCustom python==3.10.9

# 2. Activate the environment
conda activate HunyuanCustom

# 3. Install PyTorch and other dependencies using conda
# For CUDA 11.8
conda install pytorch==2.4.0 torchvision==0.19.0 torchaudio==2.4.0 pytorch-cuda=11.8 -c pytorch -c nvidia
# For CUDA 12.4
conda install pytorch==2.4.0 torchvision==0.19.0 torchaudio==2.4.0 pytorch-cuda=12.4 -c pytorch -c nvidia

# 4. Install pip dependencies
python -m pip install -r requirements.txt
# 5. Install flash attention v2 for acceleration (requires CUDA 11.8 or above)
python -m pip install ninja
python -m pip install git+https://github.com/Dao-AILab/[email protected]

In case of running into float point exception(core dump) on the specific GPU type, you may try the following solutions:

# Option 1: Making sure you have installed CUDA 12.4, CUBLAS>=12.4.5.8, and CUDNN>=9.00 (or simply using our CUDA 12 docker image).
pip install nvidia-cublas-cu12==12.4.5.8
export LD_LIBRARY_PATH=/opt/conda/lib/python3.8/site-packages/nvidia/cublas/lib/

# Option 2: Forcing to explicitly use the CUDA 11.8 compiled version of Pytorch and all the other packages
pip uninstall -r requirements.txt  # uninstall all packages
pip install torch==2.4.0 --index-url https://download.pytorch.org/whl/cu118
pip install -r requirements.txt
pip install ninja
pip install git+https://github.com/Dao-AILab/[email protected]

Additionally, you can also use HunyuanVideo Docker image. Use the following command to pull and run the docker image.

# For CUDA 12.4 (updated to avoid float point exception)
docker pull hunyuanvideo/hunyuanvideo:cuda_12
docker run -itd --gpus all --init --net=host --uts=host --ipc=host --name hunyuanvideo --security-opt=seccomp=unconfined --ulimit=stack=67108864 --ulimit=memlock=-1 --privileged hunyuanvideo/hunyuanvideo:cuda_12
pip install gradio==3.39.0 diffusers==0.33.0 transformers==4.41.2

# For CUDA 11.8
docker pull hunyuanvideo/hunyuanvideo:cuda_11
docker run -itd --gpus all --init --net=host --uts=host --ipc=host --name hunyuanvideo --security-opt=seccomp=unconfined --ulimit=stack=67108864 --ulimit=memlock=-1 --privileged hunyuanvideo/hunyuanvideo:cuda_11
pip install gradio==3.39.0 diffusers==0.33.0 transformers==4.41.2

🧱 Download Pretrained Models

The details of download pretrained models are shown here.

🚀 Parallel Inference on Multiple GPUs

For example, to generate a video with 8 GPUs, you can use the following command:

Run Single-Subject Video Customization

cd HunyuanCustom

export MODEL_BASE="./models"
export PYTHONPATH=./
torchrun --nnodes=1 --nproc_per_node=8 --master_port 29605 hymm_sp/sample_batch.py \
    --ref-image './assets/images/seg_woman_01.png' \
    --pos-prompt "Realistic, High-quality. A woman is drinking coffee at a café." \
    --neg-prompt "Aerial view, aerial view, overexposed, low quality, deformation, a poor composition, bad hands, bad teeth, bad eyes, bad limbs, distortion, blurring, text, subtitles, static, picture, black border." \
    --ckpt ${MODEL_BASE}"/hunyuancustom_720P/mp_rank_00_model_states.pt" \
    --video-size 720 1280 \
    --seed 1024 \
    --sample-n-frames 129 \
    --infer-steps 30 \
    --flow-shift-eval-video 13.0 \
    --save-path './results/sp_720p'

Run Video-Driven Video Customization (Video Editing)

cd HunyuanCustom

export MODEL_BASE="./models"
export PYTHONPATH=./
torchrun --nnodes=1 --nproc_per_node=8 --master_port 29605 hymm_sp/sample_batch.py \
    --ref-image './assets/images/sed_red_panda.png' \
    --input-video './assets/input_videos/001_bg.mp4' \
    --mask-video './assets/input_videos/001_mask.mp4' \
    --expand-scale 5 \
    --video-condition \
    --pos-prompt "Realistic, High-quality. A red panda is walking on a stone road." \
    --neg-prompt "Aerial view, aerial view, overexposed, low quality, deformation, a poor composition, bad hands, bad teeth, bad eyes, bad limbs, distortion, blurring, text, subtitles, static, picture, black border." \
    --ckpt ${MODEL_BASE}"/hunyuancustom_editing_720P/mp_rank_00_model_states.pt" \
    --seed 1024 \
    --infer-steps 50 \
    --flow-shift-eval-video 5.0 \
    --save-path './results/sp_editing_720p'
    # --pose-enhance # Enable for human videos to improve pose generation quality.

Run Audio-Driven Video Customization

cd HunyuanCustom

export MODEL_BASE="./models"
export PYTHONPATH=./
torchrun --nnodes=1 --nproc_per_node=8 --master_port 29605 hymm_sp/sample_batch.py \
    --ref-image './assets/images/seg_man_01.png' \
    --input-audio './assets/audios/milk_man.mp3' \
    --audio-strength 0.8 \
    --audio-condition \
    --pos-prompt "Realistic, High-quality. In the study, a man sits at a table featuring a bottle of milk while delivering a product presentation." \
    --neg-prompt "Two people, two persons, aerial view, overexposed, low quality, deformation, a poor composition, bad hands, bad teeth, bad eyes, bad limbs, distortion, blurring, text, subtitles, static, picture, black border." \
    --ckpt ${MODEL_BASE}"/hunyuancustom_audio_720P/mp_rank_00_model_states.pt" \
    --seed 1026 \
    --video-size 720 1280 \
    --sample-n-frames 129 \
    --cfg-scale 7.5 \
    --infer-steps 30 \
    --use-deepcache 1 \
    --flow-shift-eval-video 13.0 \
    --save-path './results/sp_audio_720p'

🔑 Single-gpu Inference

For example, to generate a video with 1 GPU, you can use the following command:

cd HunyuanCustom

export MODEL_BASE="./models"
export DISABLE_SP=1
export PYTHONPATH=./
python hymm_sp/sample_gpu_poor.py \
    --ref-image './assets/images/seg_woman_01.png' \
    --pos-prompt "Realistic, High-quality. A woman is drinking coffee at a café." \
    --neg-prompt "Aerial view, aerial view, overexposed, low quality, deformation, a poor composition, bad hands, bad teeth, bad eyes, bad limbs, distortion, blurring, text, subtitles, static, picture, black border." \
    --ckpt ${MODEL_BASE}"/hunyuancustom_720P/mp_rank_00_model_states_fp8.pt" \
    --video-size 512 896 \
    --seed 1024 \
    --sample-n-frames 129 \
    --infer-steps 30 \
    --flow-shift-eval-video 13.0 \
    --save-path './results/1gpu_540p' \
    --use-fp8

Run with very low VRAM

cd HunyuanCustom

export MODEL_BASE="./models"
export CPU_OFFLOAD=1
export PYTHONPATH=./
python hymm_sp/sample_gpu_poor.py \
    --ref-image './assets/images/seg_woman_01.png' \
    --pos-prompt "Realistic, High-quality. A woman is drinking coffee at a café." \
    --neg-prompt "Aerial view, aerial view, overexposed, low quality, deformation, a poor composition, bad hands, bad teeth, bad eyes, bad limbs, distortion, blurring, text, subtitles, static, picture, black border." \
    --ckpt ${MODEL_BASE}"/hunyuancustom_720P/mp_rank_00_model_states_fp8.pt" \
    --video-size 720 1280 \
    --seed 1024 \
    --sample-n-frames 129 \
    --infer-steps 30 \
    --flow-shift-eval-video 13.0 \
    --save-path './results/cpu_720p' \
    --use-fp8 \
    --cpu-offload 

Run a Gradio Server

cd HunyuanCustom

# Single-Subject Video Customization
bash ./scripts/run_gradio.sh 

# Video-Driven Video Customization
bash ./scripts/run_gradio.sh --video

# Audio-Driven Video Customization
bash ./scripts/run_gradio.sh --audio

🔗 BibTeX

If you find HunyuanCustom useful for your research and applications, please cite using this BibTeX:

@misc{hu2025hunyuancustom,
      title={HunyuanCustom: A Multimodal-Driven Architecture for Customized Video Generation}, 
      author={Teng Hu and Zhentao Yu and Zhengguang Zhou and Sen Liang and Yuan Zhou and Qin Lin and Qinglin Lu},
      year={2025},
      eprint={2505.04512},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2505.04512}, 
}

Acknowledgements

We would like to thank the contributors to the HunyuanVideo, HunyuanVideo-Avatar, MimicMotion, SD3, FLUX, Llama, LLaVA, Xtuner, diffusers and HuggingFace repositories, for their open research and exploration.