Wan2GP 是专为“显卡资源有限”的用户打造的**开源视频生成工具箱**。它把目前最先进的文生视频、图生视频、视频编辑模型（Wan 2.1、Hunyuan Video、LTX Video 等）做了重度优化和封装，让只有 6-8 GB 显存的旧显卡也能跑得动、跑得快，并提供了一整套 Web 界面与周边工具，做到“开箱即玩”。
一句话总结：**把原本需要 20-80 GB 显存才能跑的前沿视频模型，压缩到 6-10 GB 就能用，而且还附带丰富的后期与定制功能。**
核心亮点
1. **极低门槛**
• 6 GB 显存即可跑部分模型；支持 GTX 10/20 系等老显卡。
2. **一键全能 Web 界面**
• 纯浏览器操作：选模型、填提示词、排队生成、查看结果。
• 自动根据显卡架构下载最合适的模型权重。
3. **丰富的生成与后期工具**
• Mask 编辑器、提示词增强器、Pose/Depth/光流提取器。
• 音频自动生成（MMAudio），可直接给无声视频配同步音轨。
• Film Grain、超分、补帧、首尾帧续写、视频信息提取等。
4. **强大的可控生成**
• 支持 VACE 控制网：替换人物、场景扩展、镜头运动、多人对话同时说话（Multitalk Dual Voices）。
• 最新 7.0 版把 Wan 模型扩展成“AI PS”：一次出 4 张图、图生图高清修复、Flux Kontext 图像编辑、局部重绘等。
5. **社区共享与自定义**
• 类似 LoRA 的“Finetune”机制：用户可把新模型或微调配方写成轻量定义文件，云端托管，他人一键加载。
• 官方已预置多条加速路线（CausVid、AccVideo、FusioniX 等），最快 4-8 步出片。
6. **排队与批量**
• 可把多条任务列队，关掉浏览器回来收片；支持把刚生成的视频直接当“控制视频”继续延长或做二次创作。
典型使用场景
• 个人创作者：用笔记本 3060 6 GB 就能做 720p 短片。
• 小型工作室：老显卡集群并行跑批量广告片。
• 研究者：快速验证 LoRA、ControlNet 或新模型思路，无需购置 A100/H100。
附赠10个视频提示词给你参考
1.
A tall, athletic woman with toned legs and long straight black hair, wearing a tight white bikini, walking along a private beach at golden hour. The sun casts warm light on her tanned skin and the ocean waves gently crash behind her. She turns and smiles with a confident, seductive gaze as the wind blows her hair slightly.
2.
A sultry brunette in a deep red silk robe, standing in a luxurious bedroom with velvet curtains and soft candlelight. Her hair is loose and messy, and she slowly unties the robe to reveal a black lace teddy underneath. The camera captures every subtle movement, the glow of the candles reflecting in her eyes and on her smooth skin.
3.
A young, exotic beauty with golden tan skin and long curly brown hair, wearing a translucent white shirt and high-waisted jeans, dancing in a rooftop party at sunset. The city skyline is bathed in orange and pink hues, and the breeze lifts her hair as she sways to the music. Her smile is warm, but her eyes are full of mystery.
4.
A high-fashion model with a shaved head and sharp cheekbones, wearing a futuristic silver latex bodysuit, walking through a mirrored room filled with infinite reflections. Neon lights pulse around her, and the camera follows her movements in slow motion, highlighting the curves of her body and the shine of the fabric.
5.
A vintage Hollywood-style blonde bombshell with pin-up curls and bold red lips, wearing a tight emerald green satin dress, posing in front of a classic car in 1950s style. She winks at the camera while adjusting her gloves, and the sunlight enhances the sparkle in her eyes and the glamour of the scene.
6.
A seductive woman with a tattooed back and long wavy auburn hair, wearing a black corset and leather skirt, standing in a dimly lit jazz club. The room is filled with smoky lighting and soft saxophone music. She leans against a bar, sipping a cocktail, and glances at the camera with a knowing, sultry smile.
7.
A Brazilian beauty with a perfect hourglass figure and caramel skin, wearing a colorful carnival-inspired bikini adorned with feathers and sequins, dancing on a crowded beach during Rio Carnival. The crowd cheers, confetti flies in the air, and the camera captures her sensual dance moves and radiant energy.
8.
A mysterious gothic beauty with pale skin and black lipstick, wearing a fishnet stocking and a cropped velvet choker, walking through a haunted mansion at midnight. Candles flicker in the dark hallways, and her long red coat flows behind her. Her eyes are sharp and hypnotic as she turns to look at the camera.
9.
A confident Asian woman with glossy black hair and cat-eye makeup, wearing a tight white crop top and high-waisted shorts, walking down a crowded Tokyo street at night. Neon signs flash around her, and she moves with grace and confidence. The camera captures her silhouette against the bright city lights and the subtle sway of her hips.
10.
A seductive French woman with tousled blonde hair and natural makeup, wearing a loose white blouse and no bra, sitting on a Parisian balcony overlooking the city at dusk. She lights a cigarette and takes a slow drag, the breeze gently blowing her hair. The camera lingers on her lips, the soft lighting, and the romantic Parisian skyline behind her.
整合包说明
1 支持50系，最低可以支持到英伟达6G显卡，安装好自己显卡能支持的cuda最高版本
2 未修改任何代码。
3 代码和模型是分开打包的，只下载整合包的话打开启动后会自动下载模型文件。ckpts和loras_i2v是模型文件夹。我已经默认下载了几个模型文件。下载到整合包ckpts里面就可以了。
因为模型文件实在是太大了，只上传到夸克了
https://pan.quark.cn/s/6731face8c2e
4 为了照顾那些显卡不行的小伙伴，默认的参数 选择的是LowRAM_LowVRAM, profile 4 (Default): at least 32 GB of RAM and 12 GB of VRAM, if you have little VRAM or want to generate longer videos。如果你显卡比较好，可以选择更高的参数。Configuration-Performance-Profile (for power users only, not needed to change it)
生成的视频效果演示
https://www.bilibili.com/video/BV14yhFz3Ehm/?vd_source=f0ca2a91a0d1850ea46d21a82729acaa
https://www.bilibili.com/video/BV18XWQzGES6/?vd_source=f0ca2a91a0d1850ea46d21a82729acaa
https://www.bilibili.com/video/BV1MfnYzyETM/?vd_source=f0ca2a91a0d1850ea46d21a82729acaa
https://www.bilibili.com/video/BV1knn6zXECf/?vd_source=f0ca2a91a0d1850ea46d21a82729acaa
操作教程
https://www.bilibili.com/video/BV1K682zqESs/?vd_source=f0ca2a91a0d1850ea46d21a82729acaa
Wan2GP-7.1 是汉化版
20250925 更新记录
1更新到官方版本8.73
20250929 更新记录
1更新到官方版本8.761
20251001 更新记录
1 更新到官方版本8.9
20251008 更新记录
1 更新到官方版本8.995
20251021 更新记录
1 更新到官方版本8.999
20251022 更新记录
1 更新到官方版本9.0
20251027更新记录
1更新到官方版本9.21
2 上传了部分模型文件夹到ckpts
20251102 更新记录
1更新到官方版本9.3
20251106更新记录
1更新到官方版本9.35
2 更新一些新模型到ckpts文件夹
20251110更新记录
1 更新到官方版本9.41
20251117 更新记录
1 更新到官方版本9.44
20251121 更新记录
1 更新到官方版本9.5
20251124 更新记录
1 更新到官方版本9.52
20251125更新记录
1更新到官方9.62
2 新的模型文件 已经上传到夸克那个下载地址
20251130 更新记录
1 更新到官方9.72
2 建议全新解压，要覆盖的话 除了ckpts文件夹不要删，其他都删了再覆盖
解压之后运行错误
🗝️ 检测到环境变量，尝试自动验证...
✅ 密钥有效！有效期：永久有效
🚀 验证通过，正在启动主程序...
Traceback (most recent call last):
File "app.py", line 9369, in
File "app.py", line 9322, in create_ui
File "app.py", line 7003, in generate_video_tab
File "app.py", line 2701, in setup_loras
Exception: --lora-dir should be a path to a directory that contains Loras
Autosave: Queue is empty, nothing to save.
按任意键退出...
你修改过什么？
我也有一樣的問題
解壓後直接執行
沒有修改
什么都没有改动，下载下来解压运行就出现上面的提示。
loras_i2v 这个文件夹你没有下载。
OK了!感謝!
我都一樣有這錯誤, 之前都不需要lora i2v就只需要下載wan的zip檔即可用, 現在卻不行, 解壓後用不了
那你就不要升级到最新版嘛，否则升级的意义在哪里呢？
看名字就知道啊
我已經正在下載lora i2v了, 這個新的文件夾是在應用在什麼方面的?
WanGP v8.31出来了，剑哥，出来干活啦。
已经更新到了8.4
老大，这个最低内存需要多大？ 32G能运行么
应该可以
只要选择wan2.1就页面就一堆错误。
学习交流
哥，8.1官方说支持InfiniteTalk，为啥打开页面提示好几哥错误啊？谢谢
什么错误？
Exception: --lora-dir should be a path to a directory that contains Loras
这边没办法贴图啊？Wan2.1
错误
X
Refresh
Delete
Status
Infinitetalk Single Speaker 480p
The Infinitetalk model is an improved version of Multitalk that supports very long videos. This is the single speaker version.
Attention mode auto/sage2, Data Type BF16, Quantization Scaled Int8
Apply
Save
错误
X
Images as starting points for new videos
X
错误
Prompts (each new line of prompt will generate a new video, # lines = comments, !
lines = macros)
错误
Video Info and Late Post Processing & Audio Remu
看看黑色窗口提示什么
Exception: --lora-dir should be a path to a directory that contains Loras
操作什么出现的？lora文件是否放好了？
这个要怎么用啊！看不懂啊！
仔细看文章说明 有操作说明
您好，请问这个的缓存文件在哪个文件夹，谢谢
没注意
更新包怎用?
你好，看到7.77的包3个多G，是完整体还是说更新内容啊？是不是把7.77下载下来覆盖到原来的目录就行了？
完整包，只是没有模型文件
8个小时过去了，5秒图生的视频。不至于吧，剑哥帮忙看看
我怎么看呢？你啥都不说 比如电脑配置啥的？
The whole model was pinned to reserved RAM: 62 large blocks spread across 13860.55 MB
Hooked to model 'transformer2' (WanModel)
48%|████████████████████████████████████▌ | 19/40 [2:23:58
这里面有支持NSFW的模型嘛？
你试试不就知道了？我没试过
剑佬，你这个包torch是什么版本的？有没有SageAttention加速？SageAttention是什么版本的？
accelerate 1.9.0
addict 2.4.0
aiofiles 23.2.1
aiohappyeyeballs 2.6.1
aiohttp 3.12.14
aiosignal 1.4.0
alembic 1.16.4
altgraph 0.17.4
annotated-types 0.7.0
antlr4-python3-runtime 4.9.3
anyio 4.9.0
asteroid-filterbanks 0.4.0
astunparse 1.6.3
async-timeout 5.0.1
attrs 25.3.0
audioread 3.0.1
auto-py-to-exe 2.46.0
av 15.0.0
bottle 0.13.4
bottle-websocket 0.2.9
certifi 2025.7.14
cffi 1.17.1
charset-normalizer 3.4.2
click 8.2.1
cloudpickle 3.1.1
colorama 0.4.6
coloredlogs 15.0.1
colorlog 6.9.0
contourpy 1.3.2
cryptography 45.0.5
cycler 0.12.1
dashscope 1.23.9
decorator 4.4.2
decord 0.6.0
diffusers 0.34.0
docopt 0.6.2
easydict 1.13
Eel 0.18.2
einops 0.8.1
exceptiongroup 1.3.0
fastapi 0.116.1
ffmpeg-python 0.2.0
ffmpy 0.6.0
filelock 3.13.1
flash_attn 2.7.4.post1
flatbuffers 25.2.10
fonttools 4.59.0
frozenlist 1.7.0
fsspec 2024.6.1
ftfy 6.3.1
future 1.0.0
gevent 25.5.1
gevent-websocket 0.10.1
gradio 5.23.0
gradio_client 1.8.0
greenlet 3.2.3
groovy 0.1.2
h11 0.16.0
hf-xet 1.1.5
httpcore 1.0.9
httpx 0.28.1
huggingface-hub 0.34.3
humanfriendly 10.0
hydra-core 1.3.2
HyperPyYAML 1.2.2
idna 3.10
imageio 2.37.0
imageio-ffmpeg 0.6.0
importlib_metadata 8.7.0
importlib_resources 6.5.2
Jinja2 3.1.4
joblib 1.5.1
jsonschema 4.25.0
jsonschema-specifications 2025.4.1
julius 0.2.7
kiwisolver 1.4.8
lazy_loader 0.4
librosa 0.11.0
lightning 2.5.2
lightning-utilities 0.14.3
llvmlite 0.44.0
loguru 0.7.3
Mako 1.3.10
markdown-it-py 3.0.0
MarkupSafe 2.1.5
matplotlib 3.10.3
mdurl 0.1.2
misaki 0.9.4
mmgp 3.5.10
moviepy 2.1.2
mpmath 1.3.0
msgpack 1.1.1
multidict 6.6.3
mutagen 1.47.0
networkx 3.3
ninja 1.11.1.4
numba 0.61.2
numpy 1.26.4
nvidia-ml-py 12.575.51
omegaconf 2.3.0
onnxruntime-gpu 1.22.0
open_clip_torch 2.32.0
opencv-python 4.11.0.86
opencv-python-headless 4.11.0.86
optimum-quanto 0.2.7
optuna 4.4.0
ordered-set 4.1.0
orjson 3.11.0
packaging 25.0
pandas 2.3.1
pefile 2023.2.7
peft 0.15.0
piexif 1.1.3
pillow 11.0.0
pip 25.1
platformdirs 4.3.8
pooch 1.8.2
primePy 1.3
proglog 0.1.12
propcache 0.3.2
protobuf 6.31.1
psutil 7.0.0
pyannote.audio 3.3.2
pyannote.core 5.0.0
pyannote.database 5.1.3
pyannote.metrics 3.2.1
pyannote.pipeline 3.0.1
pyarmor 9.1.7
pyarmor.cli.core 7.6.7
pycparser 2.22
pydantic 2.10.6
pydantic_core 2.27.2
pydub 0.25.1
pygame 2.6.1
Pygments 2.19.2
pyinstaller 6.15.0
pyinstaller-hooks-contrib 2025.8
pyloudnorm 0.1.1
PyMatting 1.1.14
pynvml 12.0.0
pyparsing 3.2.3
pyreadline3 3.5.4
python-dateutil 2.9.0.post0
python-dotenv 1.1.1
python-multipart 0.0.20
pytorch-lightning 2.5.2
pytorch-metric-learning 2.8.1
pytz 2025.2
pyvers 0.1.0
pywin32-ctypes 0.2.3
PyYAML 6.0.2
referencing 0.36.2
regex 2024.11.6
rembg 2.0.65
requests 2.32.4
rich 14.0.0
rpds-py 0.26.0
ruamel.yaml 0.18.14
ruamel.yaml.clib 0.2.12
ruff 0.12.4
safehttpx 0.1.6
safetensors 0.5.3
scikit-image 0.25.2
scikit-learn 1.7.1
scipy 1.15.3
segment-anything 1.0
semantic-version 2.10.0
semver 3.0.4
sentencepiece 0.2.0
setuptools 78.1.1
shellingham 1.5.4
six 1.17.0
sniffio 1.3.1
sortedcontainers 2.4.0
sounddevice 0.5.2
soundfile 0.13.1
soxr 0.5.0.post1
speechbrain 1.0.3
SQLAlchemy 2.0.41
starlette 0.47.1
sympy 1.13.3
tabulate 0.9.0
tensorboardX 2.6.4
tensordict 0.9.1
threadpoolctl 3.6.0
tifffile 2025.5.10
timm 1.0.17
tokenizers 0.21.2
tomli 2.2.1
tomlkit 0.13.3
torch 2.7.0+cu128
torch-audiomentations 0.12.0
torch_pitch_shift 1.2.5
torchaudio 2.7.0+cu128
torchdiffeq 0.2.5
torchmetrics 1.7.4
torchvision 0.22.0+cu128
tqdm 4.67.1
transformers 4.53.1
triton 3.0.0
typer 0.16.0
typing_extensions 4.12.2
tzdata 2025.2
urllib3 2.5.0
uvicorn 0.35.0
wcwidth 0.2.13
websocket-client 1.8.0
websockets 15.0.1
wheel 0.45.1
win32_setctime 1.2.0
yarl 1.20.1
zipp 3.23.0
zope.event 5.1.1
zope.interface 7.2
zstandard 0.23.0
大佬，我想请问一下，为什么在用wan2.2文生视频时，显卡的专用内存占用率为6G，共享内存却使用了30个G呢？
使用的显卡是4070ts，显存为16G，现在按照您给出的提示词生成5秒视频运行时间大概是30分钟，其他设置都是默认的。
谢谢大佬~
因为你在设备选项那里没有选 符合自己电脑配置的选项。
请问是指configuration-performance里的选项吗
整合包说明 仔细看看
谢谢大佬，我再仔细研究研究
剑哥，这个可以加上那个加速开关吗？我看其他人的视频好像很厉害的样子啊
有这个你其他视频生成的整合包都可以删了