Differential ascii art renderer that enables unsupervised learning of image-to-ascii-art conversion.
The renderer is easily combined with typical 2D CNN models, implemented in layers.py .
 |
 |
more samples in generated_images directory.
python3 predict.py weights.h5 line_drawing_dir/*.png
- press 'w' key on window to write out the ascii art.
predict:
- tensorflow>=2.2.0
- opencv-python>=4.0.0
- numpy
train/experiment:
- tqdm
- opencv-contrib-python>=4.0.0
- pandas
- scikit-image
- fusewarp(https://github.com/stnk20/fusewarp)