Faster neural doodle
This is my try on drawing with neural networks, which is faster than Alex J. Champandard's version, and similar in quality. This approach is based on neural artistic style method (L. Gatys), whereas Alex's version uses CNN+MRF approach of Chuan Li.
It takes several minutes to redraw
Renoir example using GPU and it will easily fit in 4GB GPUs. If you were able to work with Justin Johnson's code for artistic style then this code should work for you too.
You can find even faster version here.
- torch.cudnn (optional)
- python + numpy + scipy + h5py + sklearn
Tested with python2.7 and latest
Do it yourself
First download VGG-19.
cd data/pretrained && bash download_models.sh && cd ../..
Use this script to get intermediate representations for masks.
python get_mask_hdf5.py --n_colors=4 --style_image=data/Renoir/style.png --style_mask=data/Renoir/style_mask.png --target_mask=data/Renoir/target_mask.png
Now run doodle.
th fast_neural_doodle.lua -masks_hdf5 masks.hdf5
And here is the result. First row: original, second -- result.
Processing the image at low resolution first can provide a significant speed-up. You can pass a list of resolutions to use when processing. Passing
256 means that the images and masks should be resized to
256x256 resolution. With
0 passed no resizing is done. Here is an example for cmd parameters:
-num_iterations 450,100 -resolutions 256,0Which means: work for 450 iterations at
256x256resolution and 100 iterations at original.
Renoir examples take ~1.5 min to process with these options.
You can also provide target image to use in content loss (in the same way as in neural artisctic style algorithm) via
--target_image option of
python get_mask_hdf5.py --n_colors=4 --style_image=data/Renoir/style.png --style_mask=data/Renoir/style_mask.png --target_mask=data/Renoir/creek_mask.jpg --target_image=data/Renoir/creek.jpg th fast_neural_doodle.lua -masks_hdf5 masks.hdf5
Upper left: target image. Upper right: neural doodle with target image, i.e. both the masks and content loss were used. Lower left: regular neural doodle without content loss. Lower right: stylization without masks, with high style weight, obtained via neural style code. With high style weight, stylization tends to mix unrelated parts of image, such as patches of grass floating in the sky on last picture. Neural doodle with content loss allows to generate highly stylized images without this problem.
- nn (CPU/GPU mode)
- clnn (not tested yet..)
-backend cudnndo not forget to switch
The code is heavily based on Justin Johnson's great code for artistic style.