If you like deckbuilding + roguelike progression, Tsukuyomi is a fun climb — pick your path, fight, shop, deal with weird events. Each character has their own deck style, the fights are snappy, and it’s got that modern gothic anime vibe 🎨.

The “AI Kaneko” thing — every 10 events you get a generated card based on your play — is unique. But honestly? Some cards felt like slightly re-skinned ones I’d already seen 🤷‍♀️. The whole AI art choice is bold, but risky — especially in today’s climate. Kaneko says it’s trained only on his own art, but still… people will have strong opinions.

✅ Pros:

Fun card combat + clear progression

Nice art direction + anime gothic style

Mystery events keep runs fresh

⚠ Cons:

AI card system feels hit-or-miss and not sure it bring anything to the game

Verdict: 👍 Fun if you like the genre — but know what you’re stepping into with the AI system.

這是我在玩過這款遊戲後的一些個人感想和建議
1沒必要告訴玩家選擇某項後會得到什麼，讓玩家根據自己的直覺、喜好、習慣、性格去做出屬於自己的選擇，這樣才能讓ai生成出更符合玩家本性的卡牌，當玩家為了獲取某項獎勵（不論那是什麼）而做出了與自己本意相違背的選擇，遊戲的性質就變了。這一點同樣適用於為了讓其他玩家推薦並且官方選中自己生成的隨機卡牌並將其實裝至遊戲中而非過一段時間或版本更新後就被清理掉，玩家故意做出某些帶有明顯傾向性選擇的情況。個人認為，作為一名純粹的玩家，玩遊戲還是應該先讓自己爽，而不應該過多地去糾結自己的專屬牌能否被保留或者被他人注意到從而獲得某種“榮譽”。
2在選擇參與賭博還是教訓活動舉辦者那關，當手中的金額不足時無法選擇參與，然後在之後生成專屬卡牌時認定玩家是一名正直的，嫉惡如仇的人，這是不合理的，我不選擇賭博僅僅是因為遊戲機制不讓我參加，而不是我真的抱有這樣的想法。應該讓玩家在金額不足時依然可以選擇參與賭博，但提示玩家需要支付額外的“代價”，或者失敗機率會增加（對手可能會出千）。
3我不確定，目前這款遊戲是只能根據玩家採取的行動讓ai生成專屬卡牌（走哪條路，面對某些情況選擇某種應對方式，通過對戰獎勵和商店構築卡組時的傾向等等），還是可以把對戰中具體到每一步的操作也納入生成專屬卡牌的參考標準。舉例來說，面對敵人的攻擊，我選擇有限的防禦同時攻擊對方的均衡戰術（以盡可能少的傷害，同時兼顧擊殺效率），絕對防禦或治療自己的同時給對方掛持續傷害的龜甲戰術（只要不限制回合數，只要我比對面更持久，贏的就一定是我），還是不防禦瘋狂攻擊以傷換傷的瘋狗戰術（只要是可以攻擊的牌就全部打出去，哪怕手裡有既能攻擊也能防禦的牌，對方可以造成巨大的傷害，也不留，只想盡快打死對方）……即使我的卡組看似是某一類別，但在具體的使用細節上，每個人的對戰風格也是不一樣的。如果能將這些數據也記錄下來並作為生成專屬卡牌的依據，結果將會更加精確。
These are some personal thoughts and suggestions after playing this game.
1. There's no need to tell players what they will get after making a choice. Let players make their own choices based on their intuition, preferences, habits, and personality. This way, the AI can generate cards that align with the player's nature. If players make choices against their intentions just to get a certain reward (whatever it may be), the nature of the game changes. This also applies to situations where players deliberately make choices with obvious biases to get their generated random cards recommended and officially selected for the game rather than being removed after a while or after a version update. Personally, as a casual player, playing the game should be about enjoying oneself first and shouldn't be overly concerned with whether one's exclusive cards are retained or noticed by others to gain some 'honour'.
2. When deciding whether to participate in gambling or punish the activity organiser, if the player cannot participate due to insufficient funds, it's unreasonable for the subsequent card generation to assume the player is righteous and hates evil. Not choosing to gamble simply because the game mechanics don't allow it doesn't reflect the player's true intent. Players should still be able to choose to gamble when funds are low, but be warned of an additional 'cost' or a higher chance of failure (the opponent might cheat).
3. I'm unsure whether this game generates cards based only on the player's actions (the path taken, choices in certain situations, tendencies when building a deck through rewards and shops, etc.) or if it can include every step of gameplay in generating exclusive cards. For example, when facing an enemy's attack, do I choose a balanced tactic with limited defence while attacking (minimising damage while ensuring efficiency), an absolute defence or self-healing while inflicting sustained damage on the opponent (as long as there is no turn limit and I outlast the opponent, I will win), or a reckless attack-for-damage tactic (using all attack cards, even if some can also defend, taking huge hits to try to defeat the opponent quickly)... Even if my deck seems to belong to a certain type, individual combat styles vary in detail. If these data could also be recorded and used as a reference for generating exclusive cards, the results would be even more accurate.

I gave this a play far past its """"relevance"""" solely out of morbid curiosity and because of the growing hole in my heart from the absence of new dungeon crawling megaten and classic kaneko aesthetic and design. The precious few characters and demons that are actual real illustrations from kaneko are all pretty good. But the moment you start seeing the AI demons and enemies it completely breaks the illusion. The style isn't even slightly consistent and they don't look like something kaneko would have ever made. They are as painfully generic and as ♥♥♥♥♥♥ looking as you'd assume. That they thought the audience was stupid enough to accept the blatant lie that their AI was exclusively trained on kanekos art is just sad. The dungeon crawing is automatic and not particularly engaging either, you just chose left or right sometimes.
Its actually quite a bit of a depressing play because the story immediately introduces you to a blatant expy of the yatagarasu from megami tensei, which furthers the always assumed but not confirmed idea that kaneko didn't leave atlus on good terms and actually still just wants to make megami tensei stuff.