Nara Institute of Science and Technology (NAIST), in collaboration with IMAGICA GROUP Inc. and OLM Digital, Inc., has developed a technology for automating the "painting" work, which is one of the animation production processes. developed.

 The number of anime works produced in Japan is increasing year by year.However, the number of animators has reached a plateau, and there is an urgent need to improve the efficiency and automation of production.In order to solve this problem, this research team aimed to automate the finishing of the animation production flow (the process of coloring the completed video).

 Using deep learning technology by fusing IMAGICA GROUP and OLM Digital's "animation production technology and knowledge" and NAIST's "machine learning and computer graphics / computer vision basic technology", Succeeded in developing the world's first automatic coloring technology specializing in Japanese animation production.

 In this technology, "setting table that defines the color of the character" and "picture before coloring" are input.Then, by making full use of the deep learning technology for object recognition, the area of ​​the image is extracted and automatic coloring is performed.Furthermore, by deleting the information that causes a decrease in accuracy at the stage before automatic coloring, or conversely, by adding a correction technology when the planned area is exceeded and the color is painted after automatic coloring. The coloring result was improved.

 This result is currently in the preliminary research stage, and it is said that it will further improve the accuracy, conduct a demonstration experiment at an animation production studio, and aim for practical use by 2020.

reference:[Nara Institute of Science and Technology] IMAGICA GROUP, OLM Digital, Nara Institute of Science and Technology develops automatic animation coloring technology using deep learning technology

University Journal Online Editorial Department

This is the online editorial department of the university journal.
Articles are written by editorial staff who have a high level of knowledge and interest in universities and education.