One-shot Detail Retouching [Paper] [Project website]

• Tensorflow implementation of One-shot Detail Retouching with Patch Space Neural Transformation Blending, published at ACM SIGGRAPH European Conference on Visual Media Production, 2023.
• This repo contains the main code.

Requirements:

To test our method, you first need to install the required packages, ideally in a virtual environment. I also recommend using python>=3.8. To install the packages, you can run the following:

pip install -r requirements.txt

Useful Arguments:

[--input_path]                     # Path to a 'before' image.
[--output_path]                    # Path to an 'after' image.
[--test_path]                      # Path to an 'input' image.
[--test_output_path]               # Target directory+filename for the output image.  
[--model_path]                     # Path to the saved models.
[--num_matrices, default=256]      # Number of transformation matrices to be blended.
[--num_mlp, default=1]             # Number of MLPs to be used. For the latest version of our technique, keep it 1.
[--patch_size, default=[3,3]]      # Size of image patches to be processed.
[--laplacian_level, default=5]     # Number of Laplacian levels for frequency decomposition.

Train:

python main.py --input_path=../Before.png --output_path=../After.png

Test:

python test.py --test_path=../Input.png  --test_output_path=../Output.png

Results:

Qualitative comparisons with state-of-the-art methods on different types of images

Citation

If you find our work relevant to your research, please cite:

@inproceedings{10.1145/3626495.3626499,
author = {Gokbudak, Fazilet and Oztireli, A. Cengiz},
title = {One-Shot Detail Retouching with Patch Space Neural Transformation Blending},
year = {2023},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3626495.3626499},
doi = {10.1145/3626495.3626499},
series = {CVMP '23}
}