Real-Time Hair Filtering with Convolutional Neural Networks

Roc R. Currius, Ulf Assarsson, Erik Sintorn

Abstract

Rendering of realistic-looking hair is in general still too costly to do in real-time applications, from simulating the physics to rendering the fine details required for it to look natural, including self-shadowing.

We show how an autoencoder network, that can be evaluated in real time, can be trained to filter an image of few stochastic samples, including self-shadowing, to produce a much more detailed image that takes into account real hair thickness and transparency.

Paper

The paper can be found at https://doi.org/10.1145/3522606.

You can download the author version of the paper from here.

Supplemental Files

• Video with evaluation comparison of our method: https://www.youtube.com/watch?v=nDbn9pzZM5w
• Code implementation: https://gitlab.com/ror3d/hair-filtering-cnn
• Video of the paper presentation: https://www.youtube.com/live/P-wRgPe8sDs?t=14407

Presentation

Bib. Reference

@inproceedings{RealTimeHairFilteringCNN:currius:2022,
    author = {Currius, R. R. and Assarsson, U. and Sintorn, E.},
    title = {Real-Time Hair Filtering with Convolutional Neural Networks},
    booktitle = {Proceedings of the ACM on Computer Graphics and Interactive Techniques},
    doi = {10.1145/3522606},
    year = {2022}
    issue_date = {May 2022},
    journal = {Proc. ACM Comput. Graph. Interact. Tech.},
    volume = {5},
    number = {1},
    articleno = {15},
}