智宸墨语

简体中文

English

简体中文

English

外观

本地部署图片去水印工具

约 648 字大约 2 分钟

Lama Cleaner去水印

2025-07-12

Lama Cleaner 介绍

Lama Cleaner是一款完全免费开源，而且没有分辨率限制的图片去水印、修复工具。Lama Cleaner，内置了多种AI 模型构建，功能相当的齐全。可用于快速去除图像中各种水印、物品、人物、字体、等对象，并支持老照片修复、文本替换图像内容等。

安装过程

Anaconda创建python 虚拟环境

conda create -n lama-clean python=3.11

注

python版本，3.7+ 即可，本次安装使用3.11 环境

conda activate lama-clean

pip安装lama-cleaner

pip install lama-cleaner -i https://pypi.tuna.tsinghua.edu.cn/simple

启动lama-cleaner

lama-cleaner --model=lama --device=cpu --port=9088 --gui

注

本机使用的是macbookpro m2pro处理器，所以使用cpu模式，如果是windows 且有nvidia 处理器，可以利用gpu进行处理，启动命令如下：

lama-cleaner --model=lama --device=cuda --port=9090 --gui

查看lama-cleaner界面

http://127.0.0.1:9090/

image-20250712145432324

FAQ 启动报错处理

下载模型文件失败

(lama-clean) dftshine@zhangleideMacBook-Pro lama % lama-cleaner --model=lama --device=cpu --port=9090 --gui
- Platform: macOS-13.0-arm64-arm-64bit
- Python version: 3.11.13
- torch: 2.7.1
- torchvision: 0.22.1
- Pillow: 11.3.0
- diffusers: 0.16.1
- transformers: 4.27.4
- opencv-python: 4.12.0.88
- xformers: N/A
- accelerate: N/A
- lama-cleaner: 1.2.5
- rembg: N/A
- realesrgan: N/A
- gfpgan: N/A


/Users/dftshine/anaconda3/envs/lama-clean/lib/python3.11/site-packages/lama_cleaner/model/ldm.py:272: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.
  @torch.cuda.amp.autocast()
[W712 13:22:23.481466000 init.cpp:768] Warning: nvfuser is no longer supported in torch script, use _jit_set_nvfuser_enabled is deprecated and a no-op (function operator())
Downloading: "https://github.com/Sanster/models/releases/download/add_big_lama/big-lama.pt" to /Users/dftshine/.cache/torch/hub/checkpoints/big-lama.pt
Traceback (most recent call last):
  File "/Users/dftshine/anaconda3/envs/lama-clean/lib/python3.11/urllib/request.py", line 1348, in do_open
    h.request(req.get_method(), req.selector, req.data, headers,
  File "/Users/dftshine/anaconda3/envs/lama-clean/lib/python3.11/http/client.py", line 1303, in request
    self._send_request(method, url, body, headers, encode_chunked)
  File "/Users/dftshine/anaconda3/envs/lama-clean/lib/python3.11/http/client.py", line 1349, in _send_request
    self.endheaders(body, encode_chunked=encode_chunked)
  File "/Users/dftshine/anaconda3/envs/lama-clean/lib/python3.11/http/client.py", line 1298, in endheaders
    self._send_output(message_body, encode_chunked=encode_chunked)
  File "/Users/dftshine/anaconda3/envs/lama-clean/lib/python3.11/http/client.py", line 1058, in _send_output
    self.send(msg)
  File "/Users/dftshine/anaconda3/envs/lama-clean/lib/python3.11/http/client.py", line 996, in send
    self.connect()
  File "/Users/dftshine/anaconda3/envs/lama-clean/lib/python3.11/http/client.py", line 1468, in connect
    super().connect()
  File "/Users/dftshine/anaconda3/envs/lama-clean/lib/python3.11/http/client.py", line 962, in connect
    self.sock = self._create_connection(
                ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/dftshine/anaconda3/envs/lama-clean/lib/python3.11/socket.py", line 863, in create_connection
    raise exceptions[0]
  File "/Users/dftshine/anaconda3/envs/lama-clean/lib/python3.11/socket.py", line 848, in create_connection
    sock.connect(sa)
TimeoutError: [Errno 60] Operation timed out
....
....
urllib.error.URLError: <urlopen error [Errno 60] Operation timed out>

处理方法：手动下载模型并放到指定路径

  • 手动下载模型文件

    用浏览器或下载工具（如迅雷）访问模型地址：

    https://github.com/Sanster/models/releases/download/add_big_lama/big-lama.pt

    （如果访问慢，可用百度网盘链接: https://pan.baidu.com/s/1s3lqJN6glMgj7P-YvWCoZA?pwd=uczj 提取码: uczj ）

  • 将模型放到缓存路径

    下载完成后，将 big-lama.pt 复制到 lama-cleaner 自动读取的缓存路径：

​ 你的错误信息中已提示路径：/Users/dftshine/.cache/torch/hub/checkpoints/ （如果文件夹不存在，手动创建 checkpoints 文件夹）

附：

即梦AI 生成的壁纸，使用 lama-cleaner 后的效果

hutao01_cleanup