1. Download DAZ Studio
Link: https://www.daz3d.com/.
DAZ Studio is free but you need to register to download it. It comes with some free assets, including the Genesis 8 Male and Female characters that the Autoface add-on uses as base meshes.
2. Download Blender 2.79b.
Link: https://www.blender.org/.
Autoface is an add-on for Blender 2.79. It does not run on Blender 2.8x.
3. Download the DAZ importer
Zip file: https://bitbucket.org/Diffeomorphic/import-daz/downloads/
Stable version: https://www.dropbox.com/s/0tfwk7v82116nt3/import-daz-v1.3.1-20181016.zip
Documentation: http://diffeomorphic.blogspot.com/p/daz-importer-version-13.html.
The DAZ importer is a Blender add-on that imports native DAZ Studio files (.duf, .dsf) into Blender. To install the add-on, follow the instructions at http://diffeomorphic.blogspot.com/p/installing-blender-add-on.html.
4. Download Autoface.
Repository: https://bitbucket.org/Diffeomorphic/autoface.
Zip file: https://bitbucket.org/Diffeomorphic/autoface/downloads/.
Install the Autoface add-on in the same way as the DAZ importer. This add-on uses some data files that are too large to put in a Bitbucket repository. Instead the file can be found on dropbox.
Link: https://www.dropbox.com/s/mf0pxjiryr9806z/autoface_data.zip.
Put this file in the Autoface directory and unzip it. This will create a folder named data.
5. Install OpenCV
This is quite complex, but the instructions at https://www.learnopencv.com/ are very useful.
Windows: https://www.learnopencv.com/install-opencv-4-on-windows/
Mac: https://www.learnopencv.com/install-opencv-4-on-macos/
Other operating systems: https://www.learnopencv.com/?s=install.
On Windows you start by installing the following software:
5.1. Visual Studio.
5.2. CMake.
5.3. Anaconda. I installed the Python 3 version but Python 2 will probably work too.
5.4. Git for Windows.
The instructions refer to install scripts whose location is not immediately obvious. The Windows versions can be found at https://github.com/spmallick/learnopencv/tree/master/InstallScripts/Windows-4.
6. Install dlib.
Official instructions: http://dlib.net/compile.html.
Github: https://github.com/davisking/dlib.
Grab the latest sources from github, go to the base folder of the dlib repository, and run
python setup.py install
7. Install tensorflow.
Well, I failed to do that, so I am not the one to give instructions. Instead, we will run tensorflow on Google Colaboratory, or Colab for short. Many useful packages such as OpenCV, Dlib and tensorflow are preinstalled on Colab.
8. Get a Google account and Google drive.
Link: https://www.google.com/drive/
They give you 15 Gb storage for free, which is more than enough to run the Autoface script.
9. Open the autoface notebook on Colab.
Link: https://colab.research.google.com/drive/1p_6MQhzvAs2Uiiy0UQUiK-oUH2B3NIju
10. Save the notebook on your Colab.
Sign in to Colab with your Google account and save a copy of the notebook. Make sure to save it as a Python 2 notebook.
11. Mount your Google drive
To mount the Google drive, execute the first cell in the notebook.
from google.colab import drive
drive.mount('/content/gdrive')
To run the cell, press on the arrow to the left of it. By clicking on the link you will get a long password that should be pasted into the box.
12. Install Expression-Net
Run the next cell:
%cd "/content/gdrive/My Drive/"
!git clone https://github.com/fengju514/Expression-Net.git
%cd Expression-Net
!cp ./kaffe/tensorflow/network_shape.py ./ResNet/
!cp ./kaffe/tensorflow/network_expr.py ./ResNet/
For more information about Expression-Net, see https://github.com/fengju514/Expression-Net.git
13. Download the learned deep models
ExpressionNet: https://www.dropbox.com/s/frq7u7z5kgxnz9e/Expression_Model.tar.gz
Identity shape model: https://www.dropbox.com/s/ej80o9lnj0k49qu/Shape_Model.tar.gz
FacePoseNet: https://www.dropbox.com/s/r38psbq55y2yj4f/fpn_new_model.tar.gz
Make sure that the ExpressionNet, Shape, and FacePoseNet models are stored in the folders Expression_Model, Shape_Model, and fpn_new_model respectively.
14. Copy the exprnet folder from Autoface to Google Drive
The Autoface add-on, which was downloaded in step 4, contains a directory named exprnet. Copy the content of this directory into the Expression-Net directory on Google Drive. The latter should have been created by the Colab notebook in step 12.
The exprnet directory contains the python file autoface.py, a data file input.csv, and a folder with example images.
Congratulations! All dependencies are now installed.
