conda env create -f environment.yaml
根据environment.yaml创建一个叫ldm的新环境
其中报错是需要github下载的两个模型无法获取,通过VPN单独下载
Installing pip dependencies: \ Ran pip subprocess with arguments:
['/home/****/.conda/envs/ldm/bin/python', '-m', 'pip', 'install', '-U', '-r', '/mnt/****/****/latent-diffusion-main/condaenv.ie4dsr_m.requirements.txt']
Pip subprocess output:
Obtaining taming-transformers from git+https://github.com/CompVis/taming-transformers.git@master#egg=taming-transformers (from -r /mnt/****/****/latent-diffusion-main/condaenv.ie4dsr_m.requirements.txt (line 13))
Cloning https://github.com/CompVis/taming-transformers.git (to revision master) to ./src/taming-transformers
Pip subprocess error:
ERROR: Command errored out with exit status 128: git clone -q https://github.com/CompVis/taming-transformers.git /mnt/****/****/latent-diffusion-main/src/taming-transformers Check the logs for full command output.
failed
CondaEnvException: Pip failed
#egg
后面的名字#environment.yaml
...
- -e git+https://github.com/CompVis/taming-transformers.git@master#egg=taming-transformers
- -e git+https://github.com/openai/CLIP.git@main#egg=clip
- -e .
放到对应位置后还是不能用,再
pip install taming-transformers
以及
pip install clip
安装很流畅,现在可以正常运行了
可以看到,两个从github手动安装的模型放在这个路径
/home/user/.conda/envs/ldm/lib/python3.8/site-packages/xxx(taming, clip)/
比较疑惑的是如果第一次安装出了问题,想要重新装environment.yaml时,不知道如何利用environment更新已有环境。直接用上面的口令会报下面的错,搜索了很久没有找到解决方案。
CondaValueError: prefix already exists: /home/fangfei/.conda/envs/ldm
更新 解决方案
conda env update -n environment_name -f environment.yaml
即可依据environment.yaml更新环境
如果不指定-n则更新到environment.yaml中命名的环境中
参考链接
测试的代码在./script/下
首先测试T2I功能,根据README.md
下载pre-trained weights (5.7GB)
mkdir -p models/ldm/text2img-large/
wget -O models/ldm/text2img-large/model.ckpt https://ommer-lab.com/files/latent-diffusion/nitro/txt2img-f8-large/model.ckpt
采样#1
# 官方采样指令
python scripts/txt2img.py
--prompt "a virus monster is playing guitar, oil on canvas"
--ddim_eta 0.0
--n_samples 4
--n_iter 4
--scale 5.0
--ddim_steps 50
This will save each sample individually as well as a grid of size n_iter x n_samples at the specified output location (default: outputs/txt2img-samples).
采样#2
python scripts/txt2img.py
--prompt "a painting of a virus monster playing guitar"
--outdir ./outputs/txt2img-samples
--plms
--ckpt ./models/ldm/sd-v1-4.ckpt
--ddim_steps 100
--H 512
--W 512
--seed 8
其中sd-v1-4.ckpt
在ComVis下载。按网页指引申请权限就可以获得了,很友好。
stable diffusion