Publications

Towards Flat Color Prediction for Comics (Extended Abstract)

AI for Visual Arts Workshop and Challenges (AI4VA) at ECCV 2024

Cite

Abstract

This extended abstract explores the partial automation of flat colorization in comic books, utilizing a ResNet-based approach and a Transformer-based approach with positional embeddings. Existing research in this field primarily focuses on models that fully colorize black line drawings, enabling amateurs to color comics. However, professional colorists require support that offers controllability and interactivity in the coloring process. A critical step towards this support is the prediction of colors based on previously published comic books from the same series. This study investigates the efficacy of these advanced computational models in facilitating a semi-automated coloring process that aligns with the needs of professional colorists, aiming to enhance productivity while maintaining artistic integrity.

Read the paper

Citation

@inproceedings{Verduyn_Winters_Tuytelaars_2024,
	title        = {Towards Flat Color Prediction for Comics},
	author       = {Verduyn, Marnix and Winters, Thomas and Tuytelaars, Tinne},
	year         = 2024,
	booktitle    = {AI for Visual Arts Workshop and Challenges (AI4VA) at the European Conference on Computer Vision (ECCV 2024)},
	address      = {Milan, Italy},
	url          = {https://lirias.kuleuven.be/retrieve/781880}
}

Read the paper All publications