There have been some unique uses of AI software when it comes to the creation of art, but the new project from Japanese tech company AlgoAge, Inc is one that has to be marked as game-changing.
As AlgoAge’s new DeepAnime program allows for the partial automatic animation of still image anime artwork.
We are about to get all tech, so hold on for a bit as we try to explain.
Using Generative Adversarial Networks (or GAN’s, which if it isn’t the villain’s name in some cyberpunk story, now has to be) that we have before mentioned when we covered its use in AI randomized anime character creations. Using GAN’s, data is categorized and in that process, an AI is able to create new pictures in the same style. For the Moe girl use, it was able to understand the formula that makes a cute anime character, and then create its own.
Hi, we still are not done tech nerding at you.
There was some issue when it came to trying to use the same tech to animate images, and for that animation to look natural. AlgoAge, taking on the challenge, developed a deep learning module that automatically generates animations when given an image and voice to work with.
Congrats to AlgoAge for the amazing work!
Impressive is such a mild way to put the work that AlgoAge has done, as the animations indeed do have strengths to them, currently matching what you see in many visual novels. While not going to make up an entire episode of something yet, it is a great step in that direction.
Anime NYC powered by Crunchyroll 2021 is almost here! Check out the official site to grab your ticket, learn more about exhibitors & attending guests, and what awaits you at Anime NYC!
And stay tuned to Nerdier Tides as we get you ready for it!