ubuntu安装nerfstudio


博主配置:

  • ubuntu18.04
  • gcc11 (需要$\geq$8)
  • conda
  • cuda11.8
  • 安装教程(基本来源于此,除了Dependencies加入额外内容)

#Create environment

1
2
3
> conda create --name nerfstudio -y python=3.8
> conda activate nerfstudio
> python -m pip install --upgrade pip

#Dependencies

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
> pip uninstall torch torchvision functorch tinycudann
> pip install ninja git+[https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch](https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch)


注意此处可能无法安装tiny-cuda-nn,可以先进行
> git clone --recursive https://github.com/nvlabs/tiny-cuda-nn

应该就可以了
> pip install ninja /home/ubuntu/tiny-cuda-nn-master/bindings/torch #下载路径

但是我这样子虽然成功安装,仍会报错:'MLP' object has no attribute 'layers'<br />最后还是被迫用git+方式来安装(翻墙下载),后续更新,可以尝试(我还没试过):
> git clone --recursive https://github.com/nvlabs/tiny-cuda-nn
> cd bindings/torch 

> conda activate nerfstudio  
> python setup.py install  
> 来源:[https://github.com/NVlabs/tiny-cuda-nn#pytorch-extension](https://github.com/NVlabs/tiny-cuda-nn#pytorch-extension)

#Installing nerfstudio

1
2
3
4
5
> git clone [email protected]:nerfstudio-project/nerfstudio.git
> cd nerfstudio
> pip install --upgrade pip setuptools
> pip install -e .
> ns-install-cli           (可选安装)

#Train and run viewer

1
2
3
4
5
# Download some test data: 
> ns-download-data nerfstudio --capture-name=poster 

# Train model 
> ns-train nerfacto --data data/nerfstudio/poster