Illustrious XL

Illustrious XL is an open-source, SDXL-based generative AI model created by OnomaAI Research for high-resolution anime-style art and illustration. Unlike standard SDXL models that cap native generation at 1024px, Illustrious XL trains natively at 1536x1536, delivering unprecedented sharpness and intricate detail without requiring external upscalers.

The model uses a Hybrid Prompting System that combines natural English descriptions with Danbooru2023 tags for precise creative control. It features a stable text encoder fine-tuning capability, superior anatomy understanding, and support for LoRA training and custom checkpoints. The model is deliberately left untuned to serve as a neutral base canvas for artists and developers.

Illustrius XL comes in multiple versions: v1.0 introduced the 1536px native resolution, v2.0 STABLE brought improved fine-tuning stability and predictable color science through cosine annealing, and v3.5 VPred represents the final SDXL iteration utilizing V-Prediction for maximized dynamic range. The latest version achieves deeper blacks, brighter highlights, and stronger adherence to difficult perspectives.

Released under the CreativeML OpenRAIL-M and Fair AI Public License, it is free for personal and research use but prohibits closed-source proprietary monetization. The model is fully compatible with standard AI interfaces including ComfyUI, Automatic1111, Forge, and Draw Things, with official VAEs and custom ComfyUI nodes provided by OnomaAI.