【视频生成模型】通义万相Wan2.1模型本地部署和LoRA微调

1 简介

通义万相 2.1 在 2025 年 1 月推出，2 月 25 日阿里巴巴宣布全面开源该模型。此次开源意义重大，让全球开发者都能参与到模型的应用与优化中。它提供了 14B 专业版和 1.3B 轻量版两种规格，满足不同场景需求。在权威评测 VBench 中，14B 版本以 86.22% 总分超越 Sora、Luma 等国内外模型，在运动质量、视觉质量等 14 个主要维度评测中斩获 5 项第一。1.3B 轻量版则主打高效率，在 RTX 4090 显卡上仅需 8.2GB 显存即可生成 480P 视频，4 分钟内完成 5 秒视频生成，兼容主流消费级 GPU。

本文来实测一下。

2 本地部署

2.1 配置环境

将Wan2.1工程克隆到本地：

git clone https://github.com/Wan-Video/Wan2.1.git
cd Wan2.1

安装依赖库：

# Ensure torch >= 2.4.0
pip install -r requirements.txt

如果安装flash_attn较慢，可以直接下载flash-attn安装包，离线安装，下载地址：https://github.com/Dao-AILab/flash-attention/releases

2.2 下载模型

到modelscope上下载模型：

pip install modelscope
modelscope download Wan-AI/Wan2.1-T2V-14B --local_dir ./Wan2.1-T2V-14B

3 文生视频

3.1 运行命令

Single-GPU：

python generate.py  --task t2v-14B --size 1280*720 --ckpt_dir ./Wan2.1-T2V-14B --prompt "Two anthropomorphic cats in comfy boxing gear and bright gloves fight intensely on a spotlighted stage."

如果显存较小，遇到OOM（内存不足）问题，可以使用–offload_model True和–t5_cpu选项来减少GPU内存使用。例如，在RTX 4090 GPU上：

python generate.py  --task t2v-1.3B --size 832*480 --ckpt_dir ./Wan2.1-T2V-1.3B --offload_model True --t5_cpu --sample_shift 8 --sample_guide_scale 6 --prompt "Two anthropomorphic cats in comfy boxing gear and bright gloves fight intensely on a spotlighted stage."

Multi-GPU：

pip install "xfuser>=0.4.1"
torchrun --nproc_per_node=8 generate.py --task t2v-14B --size 1280*720 --ckpt_dir ./Wan2.1-T2V-14B --dit_fsdp --t5_fsdp --ulysses_size 8 --prompt "Two anthropomorphic cats in comfy boxing gear and bright gloves fight intensely on a spotlighted stage."

3.2 生成结果

使用Wan2.1-T2V-14B模型，迭代50步：

单卡耗时约2小时，显存占用80多GB。
（和官方说的5秒视频需要10分钟不太相符，是什么原因？）

8卡耗时约15分钟，每张卡占用显存30多GB。

生成结果：
bilibili

4 图生视频

4.1 运行命令

Single-GPU:

python generate.py --task i2v-14B --size 1280*720 --ckpt_dir ./Wan2.1-I2V-14B-720P --image examples/i2v_input.JPG --prompt "Summer beach vacation style, a white cat wearing sunglasses sits on a surfboard. The fluffy-furred feline gazes directly at the camera with a relaxed expression. Blurred beach scenery forms the background featuring crystal-clear waters, distant green hills, and a blue sky dotted with white clouds. The cat assumes a naturally relaxed posture, as if savoring the sea breeze and warm sunlight. A close-up shot highlights the feline's intricate details and the refreshing atmosphere of the seaside."

Multi-GPU：

torchrun --nproc_per_node=8 generate.py --task i2v-14B --size 1280*720 --ckpt_dir ./Wan2.1-I2V-14B-720P --image examples/i2v_input.JPG --dit_fsdp --t5_fsdp --ulysses_size 8 --prompt "Summer beach vacation style, a white cat wearing sunglasses sits on a surfboard. The fluffy-furred feline gazes directly at the camera with a relaxed expression. Blurred beach scenery forms the background featuring crystal-clear waters, distant green hills, and a blue sky dotted with white clouds. The cat assumes a naturally relaxed posture, as if savoring the sea breeze and warm sunlight. A close-up shot highlights the feline's intricate details and the refreshing atmosphere of the seaside."

4.2 生成结果

8卡耗时12分钟左右，每张卡显存占用约48GB。

输入图片：

生成结果：bilibili

5 首尾帧生成视频

5.1 运行命令

Single-GPU:

python generate.py --task flf2v-14B --size 1280*720 --ckpt_dir ./Wan2.1-FLF2V-14B-720P --first_frame examples/flf2v_input_first_frame.png --last_frame examples/flf2v_input_last_frame.png --prompt "CG animation style, a small blue bird takes off from the ground, flapping its wings. The bird’s feathers are delicate, with a unique pattern on its chest. The background shows a blue sky with white clouds under bright sunshine. The camera follows the bird upward, capturing its flight and the vastness of the sky from a close-up, low-angle perspective."

Multi-GPU：

torchrun --nproc_per_node=8 generate.py --task flf2v-14B --size 1280*720 --ckpt_dir ./Wan2.1-FLF2V-14B-720P --first_frame examples/flf2v_input_first_frame.png --last_frame examples/flf2v_input_last_frame.png --dit_fsdp --t5_fsdp --ulysses_size 8 --prompt "CG animation style, a small blue bird takes off from the ground, flapping its wings. The bird’s feathers are delicate, with a unique pattern on its chest. The background shows a blue sky with white clouds under bright sunshine. The camera follows the bird upward, capturing its flight and the vastness of the sky from a close-up, low-angle perspective."

5.2 生成结果

8卡耗时30分钟左右，每张卡显存占用约48GB。

输入首帧：

输入尾帧：

生成结果：bilibili

6 提示词扩展

此外，我们还可以使用本地模型进行提示词扩展（Prompt Extension）。

对于文本到视频的任务，您可以使用Qwen/Qwen2.5-14B-Instruct、Qwen/Qwen2.5-7B-Instruct 和Qwen/Qwen2.5-3B-Instruct等模型。

对于图像到视频或最后一帧到视频的任务，您可以使用Qwen/Qwen2.5-VL-7B-Instruct和Qwen/Qwen2.5-VL-3B-Instruct等模型。

较大的模型通常提供更好的扩展结果，但需要更多的GPU内存。

可以使用参数–prompt_extension_model修改用于扩展的模型，例如：

1. 文生视频：

python generate.py  --task t2v-14B --size 1280*720 --ckpt_dir ./Wan2.1-T2V-14B --prompt "Two anthropomorphic cats in comfy boxing gear and bright gloves fight intensely on a spotlighted stage" --use_prompt_extend --prompt_extend_method 'local_qwen' --prompt_extend_target_lang 'zh'

2. 图生视频：

python generate.py --task i2v-14B --size 1280*720 --ckpt_dir ./Wan2.1-I2V-14B-720P --image examples/i2v_input.JPG --use_prompt_extend --prompt_extend_model Qwen/Qwen2.5-VL-7B-Instruct --prompt "Summer beach vacation style, a white cat wearing sunglasses sits on a surfboard. The fluffy-furred feline gazes directly at the camera with a relaxed expression. Blurred beach scenery forms the background featuring crystal-clear waters, distant green hills, and a blue sky dotted with white clouds. The cat assumes a naturally relaxed posture, as if savoring the sea breeze and warm sunlight. A close-up shot highlights the feline's intricate details and the refreshing atmosphere of the seaside."

3. 首尾帧生成视频：

python generate.py --task flf2v-14B --size 1280*720 --ckpt_dir ./Wan2.1-FLF2V-14B-720P --first_frame examples/flf2v_input_first_frame.png --last_frame examples/flf2v_input_last_frame.png --use_prompt_extend --prompt_extend_model Qwen/Qwen2.5-VL-7B-Instruct --prompt "CG animation style, a small blue bird takes off from the ground, flapping its wings. The bird’s feathers are delicate, with a unique pattern on its chest. The background shows a blue sky with white clouds under bright sunshine. The camera follows the bird upward, capturing its flight and the vastness of the sky from a close-up, low-angle perspective."

7 LoRA微调

请参考：LoRA微调Wan2.1教程

git clone https://github.com/modelscope/DiffSynth-Studio.git
cd DiffSynth-Studio
pip install -e .

Step 1: Install additional packages

pip install peft lightning pandas

Step 2: Prepare your dataset

You need to manage the training videos as follows:

data/example_dataset/
├── metadata.csv
└── train
    ├── video_00001.mp4
    └── image_00002.jpg

metadata.csv:

file_name,text
video_00001.mp4,"video description"
image_00002.jpg,"video description"

We support both images and videos. An image is treated as a single frame of video.

Step 3: Data process

CUDA_VISIBLE_DEVICES="0" python examples/wanvideo/train_wan_t2v.py \
  --task data_process \
  --dataset_path data/example_dataset \
  --output_path ./models \
  --text_encoder_path "models/Wan-AI/Wan2.1-T2V-1.3B/models_t5_umt5-xxl-enc-bf16.pth" \
  --vae_path "models/Wan-AI/Wan2.1-T2V-1.3B/Wan2.1_VAE.pth" \
  --tiled \
  --num_frames 81 \
  --height 480 \
  --width 832

After that, some cached files will be stored in the dataset folder.

data/example_dataset/
├── metadata.csv
└── train
    ├── video_00001.mp4
    ├── video_00001.mp4.tensors.pth
    ├── video_00002.mp4
    └── video_00002.mp4.tensors.pth

Step 4: Train

LoRA training:

CUDA_VISIBLE_DEVICES="0" python examples/wanvideo/train_wan_t2v.py \
  --task train \
  --train_architecture lora \
  --dataset_path data/example_dataset \
  --output_path ./models \
  --dit_path "models/Wan-AI/Wan2.1-T2V-1.3B/diffusion_pytorch_model.safetensors" \
  --steps_per_epoch 500 \
  --max_epochs 10 \
  --learning_rate 1e-4 \
  --lora_rank 16 \
  --lora_alpha 16 \
  --lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
  --accumulate_grad_batches 1 \
  --use_gradient_checkpointing

Full training:

CUDA_VISIBLE_DEVICES="0" python examples/wanvideo/train_wan_t2v.py \
  --task train \
  --train_architecture full \
  --dataset_path data/example_dataset \
  --output_path ./models \
  --dit_path "models/Wan-AI/Wan2.1-T2V-1.3B/diffusion_pytorch_model.safetensors" \
  --steps_per_epoch 500 \
  --max_epochs 10 \
  --learning_rate 1e-4 \
  --accumulate_grad_batches 1 \
  --use_gradient_checkpointing

If you wish to train the 14B model, please separate the safetensor files with a comma. For example: models/Wan-AI/Wan2.1-T2V-14B/diffusion_pytorch_model-00001-of-00006.safetensors,models/Wan-AI/Wan2.1-T2V-14B/diffusion_pytorch_model-00002-of-00006.safetensors,models/Wan-AI/Wan2.1-T2V-14B/diffusion_pytorch_model-00003-of-00006.safetensors,models/Wan-AI/Wan2.1-T2V-14B/diffusion_pytorch_model-00004-of-00006.safetensors,models/Wan-AI/Wan2.1-T2V-14B/diffusion_pytorch_model-00005-of-00006.safetensors,models/Wan-AI/Wan2.1-T2V-14B/diffusion_pytorch_model-00006-of-00006.safetensors.

If you wish to train the image-to-video model, please add an extra parameter --image_encoder_path "models/Wan-AI/Wan2.1-I2V-14B-480P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth".

For LoRA training, the Wan-1.3B-T2V model requires 16G of VRAM for processing 81 frames at 480P, while the Wan-14B-T2V model requires 60G of VRAM for the same configuration. To further reduce VRAM requirements by 20%-30%, you can include the parameter --use_gradient_checkpointing_offload.

Step 5: Test

Test LoRA:

import torch
from diffsynth import ModelManager, WanVideoPipeline, save_video, VideoData


model_manager = ModelManager(torch_dtype=torch.bfloat16, device="cpu")
model_manager.load_models([
    "models/Wan-AI/Wan2.1-T2V-1.3B/diffusion_pytorch_model.safetensors",
    "models/Wan-AI/Wan2.1-T2V-1.3B/models_t5_umt5-xxl-enc-bf16.pth",
    "models/Wan-AI/Wan2.1-T2V-1.3B/Wan2.1_VAE.pth",
])
model_manager.load_lora("models/lightning_logs/version_1/checkpoints/epoch=0-step=500.ckpt", lora_alpha=1.0)
pipe = WanVideoPipeline.from_model_manager(model_manager, device="cuda")
pipe.enable_vram_management(num_persistent_param_in_dit=None)

video = pipe(
    prompt="...",
    negative_prompt="...",
    num_inference_steps=50,
    seed=0, tiled=True
)
save_video(video, "video.mp4", fps=30, quality=5)

Test fine-tuned base model:

import torch
from diffsynth import ModelManager, WanVideoPipeline, save_video, VideoData


model_manager = ModelManager(torch_dtype=torch.bfloat16, device="cpu")
model_manager.load_models([
    "models/lightning_logs/version_1/checkpoints/epoch=0-step=500.ckpt",
    "models/Wan-AI/Wan2.1-T2V-1.3B/models_t5_umt5-xxl-enc-bf16.pth",
    "models/Wan-AI/Wan2.1-T2V-1.3B/Wan2.1_VAE.pth",
])
pipe = WanVideoPipeline.from_model_manager(model_manager, device="cuda")
pipe.enable_vram_management(num_persistent_param_in_dit=None)

video = pipe(
    prompt="...",
    negative_prompt="...",
    num_inference_steps=50,
    seed=0, tiled=True
)
save_video(video, "video.mp4", fps=30, quality=5)