cityStyle — a LoRA is inspired by City: The Animation (Kyoto Animation).

Trained exclusively on screencaps from the series, with no character embeddings. The model is trained to capture the pure art style of Kyoto Animation’s work on this project: clean, warm, and beautifully alive slice-of-life aesthetic.

Key Features:

- Signature flat shading with soft yet defined gradients

- Bold outlines — strong, expressive linework

- Warm, pleasant color palette

- Excellent at everyday scenes, city streets, parks, interiors, and the atmosphere of a modern Japanese city

- Works great for both full-body shots and portraits

Trigger Tags:

anime, cityStyle, flat shading, bold outlines

This LoRA is being released for both Anima and Flux 2 Klein 9B

Flux 2 Klein 9b Notes

The Flux 2 Klein 9B version handles stylized art very well and captures the cityStyle aesthetic beautifully. However, it can have issues with anatomy and finger count (common Flux Klein quirks).

Recommendation: Use anatomy fixers or the base model (LoRA is fully compatible) for the best possible quality.

If you want significantly better anatomy and much faster generation, try the Anima version — it has excellent anatomy (far fewer extra fingers, hands, or limbs), is extremely fast, and requires very few resources.