This is the official implementation of the SIGGRAPH Asia 2022 paper "Make Your Own Sprites: Aliasing-Aware and Cell-Controllable Pixelization". Paper can be found here or downloaded from here.
Test Pro: We launch a new pixelization method based on original repository, which can produce pixelization result from 2× to N× (N could be any integer number). Please refer to the Test Pro section below.👾👾
Please see our video demo on YouTube.
- Linux
- Python 3
- NVIDIA GPU + CUDA CuDNN
- pytorch >= 1.7.1 and torchvision >= 0.8.2
The dataset is available at https://drive.google.com/file/d/1YAjcz6lScm-Gd2C5gj3iwZOhG5092fRo/view?usp=sharing.
| Path | Description |
|---|---|
| Structure Extractor | A VGG-19 model pretrained on Multi-cell dataset. |
| AliasNet | An encoder-decoder network pretrained on Aliasing dataset. |
| I2PNet | I2PNet. |
| P2INet | P2INet. |
Please read the License before use. Unauthorized commercial use is prohibited.
My email is in my profile.
使用前请阅读License,禁止未经授权的商业使用
Create empty directory ./checkpoints/YOUR_MODEL_NAME
Put alias_net.pth and pixelart_vgg19.pth in ./
Put 160_net_G_A.pth and 160_net_G_B.pth in ./checkpoints/YOUR_MODEL_NAME
Create empty directory ./dataset/TEST_DATA/Input
Put test images in ./dataset/TEST_DATA/Input
Run the following command to test:
python test_pro.py --input ./datasets/TEST_DATA/Input --cell_size 4 --model_name YOUR_MODEL_NAME
--input could be a file or directory.
--cell_size could be any integer number.
Create empty directory ./checkpoints/YOUR_MODEL_NAME
Put alias_net.pth and pixelart_vgg19.pth in ./
Put 160_net_G_A.pth and 160_net_G_B.pth in ./checkpoints/YOUR_MODEL_NAME
Download the dataset. Create two empty directories ./datasets/TRAIN_DATA/trainA and ./datasets/TRAIN_DATA/trainB.
Put non-pixel art images in ./datasets/TRAIN_DATA/trainA and put multi-cell pixel arts in ./datasets/TRAIN_DATA/trainB.
Run the following command to train:
python train.py --gpu_ids 0 --batch_size 2 --preprocess none --dataroot ./datasets/TRAIN_DATA/ --name YOUR_MODEL_NAME
The checkpoints and logs will be saved in ./checkpoints/YOUR_MODEL_NAME.
Create empty directory ./checkpoints/YOUR_MODEL_NAME
Put alias_net.pth and pixelart_vgg19.pth in ./
Put 160_net_G_A.pth and 160_net_G_B.pth in ./checkpoints/YOUR_MODEL_NAME
Create empty directory ./dataset/TEST_DATA/Input.
Put test images in ./dataset/TEST_DATA/Input, and run python prepare_data.py to prepare data.
Run the following command to test:
python test.py --gpu_ids 0 --batch_size 1 --preprocess none --num_test 4 --epoch WHICH_EPOCH --dataroot ./datasets/TEST_DATA/ --name YOUR_MODEL_NAME
The result will be saved in ./result/YOUR_MODEL_NAME.
Software Copyright License for non-commercial scientific research purposes. Please read carefully the terms and conditions in the LICENSE file and any accompanying documentation before you download and/or use the Pixel Art and/or Non-pixel art dataset, model and software, (the "Data & Software"), including code, images, videos, textures, software, scripts, and animations. By downloading and/or using the Data & Software (including downloading, cloning, installing, and any other use of the corresponding github repository), you acknowledge that you have read these terms and conditions, understand them, and agree to be bound by them. If you do not agree with these terms and conditions, you must not download and/or use the Data & Software. Any infringement of the terms of this agreement will automatically terminate your rights under this License.
- The code adapted from pytorch-CycleGAN-and-pix2pix and SCGAN.