This is an automated animator for drawing anime/manga face using stylegan3.
- docker
- docker-compose
git clone https://github.com/FullteaOfEEIC/icon_generator.git
Save image files to input
. (Datasets and scripts for collecting datasets are not included.)
docker-compose up -d
This will automatically detect faces from images and save infos to output/output.json
.
Open your web browser and access http://localhost:11112/notebooks.
With running trim.ipynb, trimed images are saved to faces
.
(you can specify another port by editing docker-compose.yml
)
Convert images from image file to tf-records.
The following script creates tf-records to datasets
folder from images saved in faces
docker exec -it style python dataset_tool.py --source=/faces --dest=/datasets/faces.zip
docker exec -it style python train.py --outdir=/results --cfg=stylegan3-t --data=/datasets/faces.zip --gpus=2 --batch=16 --gamma=8.2 --mirror=1
for more details about options, please see https://github.com/NVlabs/stylegan3