Color mixing support system for oil paints based on learning from measured data

Ryotaro Doi

Shizuoka University

Makoto Okabe

Shizuoka University

By placing the target color in the red circle at the top of the camera image, the user can confirm which color in the prepared paint set should be mixed with which color in the base color (color in the process of mixing) placed in the green circle at the bottom of the camera image.

Abstract

We propose a color mixing support system for oil paints based on learning from measured data. A learning model is constructed to estimate “how much of each color should be mixed to reach the target color” from a set of paints prepared in advance for the base color (color in the process of mixing), and a color mixing support web application for oil paints is developed based on this model. By referring to the neural network's prediction, users can check in real time which color should be added to the color they are mixing. The results of user tests showed an average improvement trend in the accuracy of reproduction of target colors. Furthermore, the users' opinions gave a favorable impression, indicating that the system is very useful especially in “the early stages of color mixing” and “situations such as not knowing what colors to mix,” confirming its usefulness as a tool to assist in color mixing.

Paper

Master Paper

VC 2023

Video

BibTeX citation

                @masterthesis{doi.ryotaro.19_master,
                    title        = {Color mixing support system for oil paints based on learning from measured data},
                    author       = {Doi Ryotaro},
                    year         = 2025,
                    month        = {February},
                    address      = {3-5-1 Johoku, Chuo-ku, Hamamatsu City, Shizuoka Prefecture},
                    school       = {Shizuoka University},
                    type         = {Master's thesis}
                }