Legendary artist Kazuma Kaneko, known for creating the art style of Shin Megami Tensei and Persona, has tapped into his roots to bring his versions of Devil May Cry 5’s Dante, Nero, and Vergil to the roguelike deck-building game Kazuma Kaneko's Tsukuyomi, exclusively for the Nintendo Switch edition.
These three iconic Devil Hunters crop up midway through the Switch game, and if defeated, they join you as ‘Jinma’ card allies, each with unique effects mirroring their original traits, such as Dante’s Stylish Rank or Vergil’s Concentration. Kazuma Kaneko designed the original Devil Trigger forms for Dante and Vergil in Devil May Cry 3, so he’s dipping into his back catalogue here.
It made me think, there’s a version of Kazuma Kaneko’s work that feels ‘slightly louder than the original’, a bit pushed, a little more exaggerated, and that’s how he describes working with AI on the PC the game, where the system effectively reflects his style back at him as a “more extreme version” of what he’d normally create.
He compares it, loosely, to watching a performance of yourself, something that mimics your own work but leans into it harder, and that idea – AI as mimic rather than replacement – sits at the centre of Tsukuyomi, especially now the game is preparing for release on Nintendo Switch with a slightly different approach.
The Switch version changes AI's role entirely. The model itself isn’t included this time; instead, the game pulls from a curated set of the strongest AI-assisted artwork generated from the PC version, effectively freezing that experimentation into a finished form. It means players aren’t interacting with an active AI system, but they are still seeing its influence, just distilled, selected, and folded back into a more traditional pipeline.
Which, in a way, mirrors how Kaneko says he works anyway: his own process isn’t about endless iteration on paper; it’s more internal than that. He describes building characters as concepts first in his mind – thinking in terms of elements, themes, associations – before committing anything to the page. He doesn’t produce “10 or 20” drafts but instead narrows his ideas down mentally and then draws a near-final character.
Kazuma Kaneko has been open about training a model on his work being “challenging,” not just technically but creatively, especially given how wide his style is across different projects, which makes it harder for a system to pin down. But as the process evolved, the friction gave way to something more useful, with the AI beginning to suggest variations and combinations that sat slightly outside his usual instincts.
This ‘art’ is not necessarily better, not necessarily usable every time, but different, and that difference is where Kazuma Kaneko says he found value, describing the experience as a kind of “discovery,” where the system could surface ideas that go beyond what he’d normally arrive at through his own, more direct workflow.
Still, for all that experimentation, the foundation of his character design hasn’t shifted. He pushes back on the idea that his work is driven by logic or systems, saying it’s less about rules and more about “feeling,” especially when balancing the tone of a character, whether something should be unsettling, beautiful, or unexpectedly soft.
You can see that clearly in creatures like the Hariko Inu, a small, dog-like monster Kazuma Kaneko tells me is his favourite in the game, something intentionally “very, very cute” in a world filled with darker designs, and even mechanically unusual, growing stronger over time while retaining that harmless appearance. It’s that mix – cute and strange, simple but off – that defines his work.
Kazuma Kaneko has says that part of the goal of the original PC version of Tsukuyomi was to let fans connect more directly with how his ideas take shape, to “share that same feeling” of creation rather than just seeing the end result, and while the Switch version steps back from the live AI model, it still carries that intent, because what is on the Switch version is a curated roster of that art.
As an experiment, where authorship, the original Tsukuyomi is an interesting exercise, with the art created by Kazuma Kaneko in ‘collaboration’ with an AI reflecting his own style back at him, in a hyper-real sense and within the game, it’s seen as ‘inferior’ to the original artist’s work. When Kaneko talks about AI creating a “more extreme version” of his art, it doesn’t land as a warning so much as a description of what happens when an artist lets something echo their work back at them and then chooses what to keep.
And all of that is interesting, but it's telling that the real big news here is that Kazuma Kaneko’s take on Dante, Nero, and Vergil is the headline news, proving the human, artistic eye is always better than AI. No one’s playing Tsukuyomi on Switch for an AI’s version of DMC’s iconic characters, but they do want to see Kazuma Kaneko’s take on these iconic characters.
Kazuma Kaneko's Tsukuyomi released for Nintendo Switch 24 April.
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.
