Renowned Japanese artist Kazuma Kaneko, celebrated for his intricate designs in the Persona and Shin Megami Tensei series, recently experimented with artificial intelligence to replicate his distinctive art style. In his latest project, he trained an AI model to generate artwork, only to discover that the process was less efficient than his traditional methods.
The Artist Behind Iconic Demons
Kazuma Kaneko has left an indelible mark on the JRPG genre through his work at Atlus. His monster designs, blending horror, mythology, and surrealism, have become synonymous with the dark allure of Shin Megami Tensei and the stylish urban fantasy of Persona. After departing Atlus in 2023, Kaneko joined mobile game developer Colopl, signaling a new chapter in his career.
His portfolio includes hundreds of unique creatures that challenge conventional aesthetics. These designs often feature grotesque yet elegant forms, drawing from global mythologies reimagined through a Japanese lens. Fans appreciate how Kaneko’s art elevates gameplay, making battles against demons visually unforgettable.
A New Venture: Tsukuyomi: The Divine Hunter
Kaneko’s first collaboration with Colopl introduces Tsukuyomi: The Divine Hunter, a dungeon-crawling roguelike set in a mysterious world. Players ascend the Steel Tower, engaging in strategic combats using “Jinma cards” – collectible items representing powerful entities.
The game’s core loop emphasizes replayability. Procedural generation ensures varied runs, while player choices influence outcomes. Helping NPCs or aggressive combat styles shape your journey, directly impacting card acquisitions and progression.
- Dungeon exploration with roguelike elements for high replay value.
- Card-based battles requiring tactical deck-building.
- Dynamic storytelling tied to moral decisions.
Innovating with the AI Kaneko System
At the heart of Tsukuyomi lies the AI Kaneko System, a novel feature generating personalized Jinma cards. Based on in-game actions like battle performance or NPC interactions, the AI produces unique artwork mimicking Kaneko’s style.
Players can share these creations online. Popular submissions may become official cards in future updates, fostering community involvement. This mechanic aims to blend player agency with artistic expression, making every playthrough visually distinct.
Training the AI: Process and Challenges
To build the AI Kaneko System, developers fed the model thousands of Kaneko’s post-Atlus illustrations. Producer Yusuke Saito oversaw iterations, selecting outputs closest to the artist’s vision for further training. The goal was to create tens of thousands of variations efficiently.
Human characters proved straightforward for the AI, rendering lifelike figures with Kaneko’s signature flair. However, fantastical monsters – Kaneko’s specialty – posed significant hurdles. Imaginary beings like two-headed creatures resisted accurate replication.
Kaneko reflected: “It’s difficult to get it to draw creatures that don’t actually exist.” He often merged multiple AI outputs manually to approximate his ideas, but ultimately found direct drawing faster.
Why AI Fell Short for Monster Design
Kaneko’s monsters demand precision in anatomy, texture, and atmosphere. AI struggled with coherence in non-existent forms, producing inconsistencies in proportions or details. While human depictions benefited from real-world references, mythical designs required nuanced creativity beyond current models.
This revelation underscores AI’s limitations in original concept art. Training involved extensive refinement, exceeding Kaneko’s manual workflow time. He noted that starting from scratch himself yielded superior results quicker.
| Aspect | AI Generation | Manual Drawing |
|---|---|---|
| Speed for Humans | Fast | Moderate |
| Monster Accuracy | Low | High |
| Total Time | Longer | Shorter |
| Creativity | Derivative | Original |
Implications for Game Development
Kaneko’s experience highlights a broader tension in gaming: AI’s promise versus practical realities. Tools like Stable Diffusion or Midjourney accelerate prototyping, but fine art demands human intuition. Developers must weigh efficiency gains against quality trade-offs.
In mobile gaming, where rapid iteration is key, AI could shine for assets like backgrounds. Yet for signature styles like Kaneko’s, it risks diluting brand identity. Colopl’s approach innovates by tying AI to gameplay, not replacing artists outright.
Community Reactions and Future Prospects
Gaming communities buzz with mixed feelings. Enthusiasts praise the personalization in Tsukuyomi, seeing AI as a fun gimmick. Critics worry about job displacement for artists, echoing debates in industries like film and publishing.
Looking ahead, advancements in diffusion models may address Kaneko’s pain points. Multimodal training with sketches or 3D models could enhance fantasy outputs. Still, his candid assessment tempers hype, reminding us AI augments, not supplants, human talent.
Balancing Technology and Tradition in JRPGs
JRPGs thrive on evocative visuals. From Final Fantasy’s photorealism to Tales of’s anime charm, art defines immersion. Kaneko’s saga illustrates how legends adapt to tech without compromising essence.
Colopl’s game arrives amid a mobile RPG boom, competing with gacha giants. Tsukuyomi differentiates via roguelike depth and AI interactivity, potentially carving a niche.
Broader AI Trends in Creative Industries
Beyond gaming, AI reshapes illustration. Tools generate book covers or comics swiftly, but pros like Kaneko emphasize oversight needs. Ethical concerns – training data copyrights, style mimicry – persist, prompting calls for regulation.
In Japan, where manga and anime dominate, artists experiment cautiously. Kaneko’s transparency fosters dialogue, guiding ethical AI adoption.
FAQs
What is Tsukuyomi: The Divine Hunter?
A mobile roguelike RPG by Colopl featuring Kazuma Kaneko’s art, with card battles in a tower-climbing adventure.
How does the AI Kaneko System work?
It generates Jinma cards based on player actions, using AI trained on Kaneko’s style for unique artwork.
Why did Kaneko prefer drawing manually?
AI excelled at humans but faltered on monsters, making the process longer than his direct approach.
Will AI replace game artists?
Unlikely soon; Kaneko’s case shows it supplements but struggles with complex originality.
When is the game releasing?
Details pending, but it’s Colopl’s first with Kaneko post-Atlus.
Key Takeaways
- Kaneko’s AI trial prioritized monsters but highlighted tech gaps.
- Tsukuyomi innovates with player-driven art generation.
- Human creativity remains irreplaceable for bespoke designs.
References
- A Persona and Shin Megami Tensei artist trained an AI model to draw like him — GamesRadar+. 2025-05-05. https://www.gamesradar.com/games/rpg/a-persona-and-shin-megami-tensei-artist-trained-an-ai-model-to-draw-like-him-and-look-at-that-it-wasnt-worth-it/
- Persona art director and creator of infamous penis demon — GamesRadar+. No specific date. https://www.gamesradar.com/games/rpg/persona-art-director-and-creator-of-infamous-penis-demon-is-making-a-console-game-that-will-demonstrate-his-perspective-on-the-world/
- Ubisoft makes a big bet on generative AI-driven gameplay — GamesRadar+. No specific date. https://www.gamesradar.com/games/ubisoft-makes-a-big-bet-on-generative-ai-driven-gameplay-leading-to-more-interactive-and-engaging-games-ceo-says-this-work-reflects-the-direction-we-want-to-take/