Skip to content

Latest commit

 

History

History
128 lines (84 loc) · 4.13 KB

Setup.md

File metadata and controls

128 lines (84 loc) · 4.13 KB

Setup

Pytti-Tools can be run without any complex setup -- completely for free! -- via google colab. The instructions below are for users who would like to install pytti-tools locally. If you would like to use pytti-tools on google colab, click this button to open the colab notebook: Open In Colab

Requirements

  • Python 3.x
  • Pytorch
  • CUDA-capable GPU
  • OpenCV
  • ffmpeg
  • Python Image Library (PIL/pillow)
  • git - simplifies downloading code and keeping it up to date
  • gdown - simplifies downloading pretrained models
  • jupyter - (Optional) Notebook interface

The following instructions assume local setup. Most of it is just setting up a local ML environment that has similar tools installed as google colab.

1. Install git and python (anaconda is recommended)

2. Clone the pytti-notebook project and change directory into it.

The pytti-notebook folder will be our root directory for the rest of the setup sequence.

git clone https://github.com/pytti-tools/pytti-notebook
cd pytti-notebook

3. Create and acivate a new environment

conda create -n pytti-tools
conda activate pytti-tools

The environment name shows up at the beginning of the line in the terminal. After running this command, it should have changed from (base) to (pytti-tools). The installation steps that follow will now install into our new "pytti-tools" environment only.

4. Install Pytorch

Follow the installation steps for installing pytorch with CUDA/GPU support here: https://pytorch.org/get-started/locally/ . For windows with anaconda:

conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch

5. Install tensorflow

conda install tensorflow-gpu

6. Install OpenCV

conda install -c conda-forge opencv

7. Install the Python Image Library (aka pillow/PIL)

conda install -c conda-forge pillow

8. ... More conda installations

conda install -c conda-forge imageio
conda install -c conda-forge pytorch-lightning
conda install -c conda-forge kornia
conda install -c huggingface transformers
conda install scikit-learn pandas

9. Install pip dependencies

pip install jupyter gdown loguru einops seaborn PyGLM ftfy regex tqdm hydra-core adjustText exrex matplotlib-label-lines

10. Download pytti-core

  git clone --recurse-submodules -j8 https://github.com/pytti-tools/pytti-core

11. Install pytti-core

pip install ./pytti-core/vendor/AdaBins
pip install ./pytti-core/vendor/CLIP
pip install ./pytti-core/vendor/GMA
pip install ./pytti-core/vendor/taming-transformers
pip install ./pytti-core

12. (optional) Build local configs

If you skip this step, PyTTI will do it for you anyway the first time you import it.

python -m pytti.warmup

Your local directory structure probably looks something like this now:

        ├── pytti-notebook
        │   ├── config
        │   └── pytti-core

If you want to "factory reset" your default.yaml, just delete the config folder and run the warmup command above to rebuild it with PyTTI's shipped defaults.

Uninstalling and/or Updating

1. Uninstall PyTTI

pip uninstall -y ./pytti-core/vendor/AdaBins
pip uninstall -y ./pytti-core/vendor/CLIP
pip uninstall -y ./pytti-core/vendor/GMA
pip uninstall -y ./pytti-core/vendor/taming-transformers
pip uninstall -y pyttitools-core;

2. Delete PyTTI and any remaining build artifacts from installing it

rm -rf build
rm -rf config
rm -rf pytti-core

3. Downloaded the latest pytti-core and re-install

git clone --recurse-submodules -j8 https://github.com/pytti-tools/pytti-core

pip install ./pytti-core/vendor/AdaBins
pip install ./pytti-core/vendor/CLIP
pip install ./pytti-core/vendor/GMA
pip install ./pytti-core/vendor/taming-transformers
pip install ./pytti-core

python -m pytti.warmup