Qing

Downloading is trigged here: https://github.com/Sanster/lama-cleaner/blob/main/lama_cleaner/app/src/components/Editor/Editor.tsx#L1059 and using this function https://github.com/Sanster/lama-cleaner/blob/main/lama_cleaner/app/src/utils.ts#L25

pip install lama-cleaner

Probably because of fp16 precision, I have released a new version that allows you to add the `--no-half` option to the startup command, please upgrade to 0.30.1(for one-click installer, rerun...

> Thanks very much for your reply. I have reinstalled and made the changes as suggested. When doing the "Dream" process I am getting a "Something went wrong: TypeError: Failed...

1. Copy `C:\Users\myname.cache\huggingface` to a drive(to avoid redownload model after change `HF_HOME`) 1. Change `win_start.bat` file, set new directory ```bash set TORCH_HOME=your_directory set HF_HOME=your_directory @call invoke start ```

If you are using the window installer, you can try change [win_config.bat file](https://github.com/Sanster/lama-cleaner/blob/main/scripts/user_scripts/win_config.bat#L10), and rerun `win_config.bat` ``` conda install -y cudatoolkit=11.6 -c conda-forge -c nvidia pip install torch==1.13.0 --extra-index-url https://download.pytorch.org/whl/cu116...

 You need a larger memory GPU to run paint_by_exmaple, I tested on 3080, and for a 512x512 image, paint_by_example needs around 5.6GB GPU memory.

with new version(0.33.0) paint_by_example needs 2.6G (image size 512x512). Please check https://lama-cleaner-docs.vercel.app/memory#paint-by-example Good luck

Something went wrong in the model downloading process, please delete model and run win_start.bat again. If you are in China a VPN proxy might be helpful for downloading process

是的（漏了个 --model），先要升级到 0.32.0（执行 win_config.bat） ```bash c.run( f"lama-cleaner --model-dir xxxxx --model {model} --device {device} --port {port} --gui --gui-size 1400 900 --sd-cpu-textencoder" ) ```