Deep learning is a kind of machine learning method that is used widely across the world in a number of applications. Instead of working from task-specific algorithms, deep learning uses data representations instead, learning as it goes along.
And now, a collaboration of researchers from IMAGICA Group Inc., OLM Digital Inc., and the Nara Institute of Science and Technology (NAIST) have developed a deep learning technique that enables the automatic colorization of anime.
Anime is very big in many far Eastern countries, particularly Japan. And while the number of animated works has risen steadily over the past few years, the number of animators has remained pretty stable.
To try and improve the efficiency of animation, and to include more automation within the industry, the researchers concentrated on the possibility of automating the colorization of anime. And together, the succeeded in developing the world’s first automatic colorization technique for Japanese anime production.
It’s a new technique that’s based on deep learning and presented at SIGGRAPH ASIA 2018 on Dec 4-7. This is an international conference based on computer graphics and interactive techniques and is likely to attract a large crowd. Although the technique is still in its early stages of development, the research team will continue to improve its accuracy until it’s expected to launch sometime during 2020.