Japanese game developer Level-5 explained how it uses generative AI like Stable Diffusion

Updated 5 months ago on June 10, 2024

Level-5, the Japanese developer known for Ni no Kuni and Professor Layton games, has revealed that it is using generative AI to develop and promote games. In a slideshow document presented to the Japanese government's study group on Intellectual Property Rights in the Age of AI, the developer gave examples of how AI is being used.

Level-5's use of AI to generate title screen inspiration for Megaton Musashi Wired

According to a report from ITmedia, Level-5 has demonstrated how they are using the Stable Diffusion AI model to transform text into images. For Megaton Musashi: Wired, which is scheduled for release next year, the AI generated several rough mockups for the title screen, which the studio's illustrators used as a reference to create the final title screen that matches the game's concept. For the promotional images, Level-5 not only uses Stable Diffusion to generate initial layout ideas, but also applies AI (SwinIR) to sharpen the image.

Level-5 using AI to create draft for Inazuma Eleven promotional image
Level-5 demonstrate AI for video game promotional image enhancement

Level-5 is also known for their Yo-kai Watch games based on the popular children's anime. They used a stable diffusion method to create 3D images of the main character Nathan Adams/Keita Amano. The AI learned from many 2D images of the character in past games, creating different 3D images with different lighting, facial expressions, and backgrounds. The developer uses these as a guide to create the right atmosphere in the game.

For the soccer RPG Inazuma Eleven: Victory Road, Level-5 used portions of artificial intelligence images to create sections of the game's backgrounds, such as large crowds of people and groups of buildings. They then combined these AI segments with backgrounds and character illustrations created by artists.

Level-5 used Stable Diffusion AI to create 3D models of Yo-kai Watch main character Keita Amano
Inazuma Eleven backgrounds incorporate AI generated elements

In addition to image generation, Level-5 uses other types of AI. The company uses ChatGPT to create characters and the text-to-speech synthesizer VOICEVOX to create temporary voices for characters, and for game promotion, the Github programmer assistance tool Copilot and Genie AI (ChatGPT) to create code and improve game descriptions to create websites more efficiently.

The use of artificial intelligence in the creative industries continues to be a controversial topic: it has been criticized for being "soulless" and concerns that it could be used to replace human talent. Level-5, however, seems to be focused on the idea that humans will always direct, rearrange and refine what AI does, using it purely as a support tool rather than as a way to replace humans. In its transparent presentation document, the Japanese developer showed that AI can be an effective tool for "creating basic data," "making suggestions," and "improving quality."

Let's get in touch!

Please feel free to send us a message through the contact form.

Drop us a line at mailrequest@nosota.com / Give us a call over skypenosota.skype