Stippling

CSCI 2240 Final Project for the Spring 2024 by Stippling Studio (Chengfan Li, Wendi Liao and Yixuan Liu)

Special thanks to the papers “Weighted Linde-Buzo-Gray Stippling” and “Multi-class Inverted Stippling” that inspired this project.

[Demo Video]

Overview

Based on the research and exploration of stippling, we have implemented the following features.

• Weighted LBG algorithm based stippling that can convert grayscale image to stippling.

• Visualization of Voronoi diagrams, as well as splitting and merging of points.

• Video stippling with low noise and minimal flickering using WLGB.

• Multi-class Inverted Stippling for grayscale image.

• Color Stippling with palette extraction.

• UNet-based reconstruction model to restore stippling to grayscale images.

WLBG

Voronoi Diagram

Voronoi diagrams can be used to create a variety of amazing visual effects including stippling. In this project, we use jc_voronoi to calculate voronoi diagram efficiently and assign pixels to voronoi cell based on edges and site position.

Video Stippling

We haven't integrated video stippling on the master branch yet, but there are some helpful python scripts included in the [VideoStippling](./VideoStippling) folder.

MLBG

Color Stippling

Reconstruction

Data

Pre-trained model: download from gdrive and put into checkpoints folder.

Processed data for training: download from gdrive and put into data folder.

Raw data: contains all color images and grayscale images without preprocessing.

Results

Use this link to download the results of all images in the test set.

Stippling-Reconstruction Iterations

Here is the result obtained after using our model to continuously perform the Stippling-Reconstruction cycle on the grayscale image. Detailed results can be found here.

Gallery

References

Deussen, Oliver, Marc Spicker, and Qian Zheng. “Weighted linde-buzo-gray stippling.” ACM Transactions on Graphics (TOG) 36.6 (2017): 1-12.

Christoph Schulz, Kin Chung Kwan, Michael Becher, Daniel Baumgartner, Guido Reina, Oliver Deussen, and Daniel Weiskopf. 2021. Multi-Class Inverted Stippling. ACM Trans. Graph. 40, 6 (2021)

Tan, J., Echevarria, J. and Gingold, Y., 2018. Efficient palette-based decomposition and recoloring of images via RGBXY-space geometry. ACM Transactions on Graphics (TOG), 37(6), pp.1-10.