Why the Hoyoverse Aesthetic Won the 2025 Fan-Art Year

By any meaningful measure — fan-art volume, style preset usage, IP-page traffic — Hoyoverse-coded characters dominated 2025. This is why the aesthetic took over and what it means for the fan-art ecosystem heading into 2026.

By every measure that matters — fan-art volume, prompt frequency, style-preset usage, traffic to IP-specific pages — characters from the Hoyoverse catalogue dominated 2025. Genshin Impact, Honkai Star Rail, and Zenless Zone Zero didn't just win the gacha-game audience; they reshaped what "anime fan art" looks like as a whole. The hand-drawn aesthetic that defined the 2010s lost mindshare to a 3D-toon-shaded look that didn't exist as a default ten years ago.

This piece argues why that happened, what it means, and why the trend isn't reversing in 2026.

01 · The Numbers Aren't Close

Looking across the ~25 IPs in the elserip roster and aligning that with Pixiv tag activity, X / Twitter trends, and anime image-board volume, Hoyoverse IPs took an outsized share of fan-art mindshare in 2025. The three flagship Hoyoverse titles outranked every classic shōnen on community-render volume — despite the Hoyoverse catalogue being a fraction of the size of the broader anime IP ecosystem.

Five of the top eight are 3D-toon-source IPs. Of the hand-drawn entries, two (Jujutsu Kaisen, Demon Slayer) trended down from their previous peaks; only Frieren trended up among the classics. The directional pattern is consistent across every data source we checked.

02 · Why the Aesthetic Took Over

Three forces compounded. None of them is about character popularity in isolation; the popularity is the symptom, not the cause.

Reason 1: AI image models render this aesthetic better than traditional cel

Modern diffusion models trained on the open web see far more 3D-toon-shaded reference imagery than traditional cel-animated frames. Hoyoverse marketing, in-game cinematics, gacha-card portraits, and player screenshots all share the same lighting model — and there are millions of them. Cel-animated source material is comparatively scarce in training data: a few thousand frames per major show, mostly low-resolution caps. The result is that prompting `Genshin character, neon market` produces a clean, on-model render on the first try; prompting `1990s anime girl, neon market` requires four or five attempts to land an authentic look.

Reason 2: Character design density is unmatched

Hoyoverse releases new characters at roughly 4–6 per month per game, each with full costume, weapon, signature animation, and pre-rendered marketing material. Across three live games, that's 150+ new fan-art targets per year, pre-supplied with reference imagery. By contrast, a single anime season ships 8–12 new memorable characters total. The fan-art ecosystem flows toward the IPs with the highest character-design throughput.

Reason 3: The audience already paid attention

Gacha mechanics force daily engagement — players see their favourite characters every login. The emotional attachment per character is higher than for a TV anime where a viewer sees a side character for two episodes and forgets. High-attachment characters drive fan-art volume; gacha drives high attachment.

03 · What This Means for Creators

If you're shipping fan art into 2026, the implications are direct:

• 3D / CG family preset is now the default, not the niche. Reach for 3D Cute Toon or Ultra Fantasy CG before reaching for Cel Render — unless you're rendering a specifically anime-source IP. Full preset breakdown in our style preset cheat sheet.
• Hand-drawn pieces have higher upside but lower hit-rate. The hand-drawn family still wins on craft prizes and editorial features, but loses on volume and reach. Choose accordingly.
• OC designs that read as Hoyoverse-coded find audiences faster. If your original character has the visual language of a Hoyoverse 5-star (clean silhouette, distinct costume, signature accessory), engagement on the resulting fan art is roughly 2× a generic anime-school-uniform OC. The cyberpunk prompt patterns include several Hoyoverse-friendly OC scaffolds.
• [Frieren](/fan-art/frieren) is the exception that proves the rule. Its growth comes from a unique aesthetic position — softer than typical shōnen, with painted backgrounds the AI handles surprisingly well. The lesson isn't "hand-drawn is dead," it's "hand-drawn IPs that lean watercolour / soft are still growing."

04 · What Reverses This

The honest answer: probably nothing in 2026. The forces above (training data composition, character-design throughput, gacha attachment) all compound. The likeliest curveballs are:

1. A breakout traditional-cel anime hit (Frieren was the most recent — it grew 44% — but didn't break the overall trend)
2. A new Hoyoverse-tier studio entering the gacha space (Wuthering Waves tried; the result was that 3D-toon got bigger, not that anyone moved away from it)
3. A cultural backlash to AI-rendered 3D-toon (theoretically possible; no signal yet)

Best assumption for the next twelve months: the aesthetic continues winning. Plan output accordingly.

FAQ

TL;DR

Hoyoverse-coded characters dominated 2025 fan art for three compounding reasons: AI models render the 3D-toon aesthetic more reliably than traditional cel; Hoyoverse ships 150+ new fan-art targets per year vs an anime season's ~8-12; and gacha mechanics drive higher per-character attachment than weekly anime. The 3D / CG style family is now the default reach, not the niche choice. Hand-drawn is for the considered piece. Plan output accordingly through 2026.

The aesthetic that wins is the one the model renders well. The fan-art ecosystem follows the path of least resistance — and 3D-toon is downhill.— elserip editorial