Native High-Resolution Synthesis
Generates flawlessly at 1536x1536 and non-standard ratios like 1248x1824 without repeating subjects or breaking compositions.
Pollo AI: Free All-in-One AI Image and Video Generator
Try for Free
Generate Breathtaking High-Res Anime Art Natively
Stop settling for upscaled distortions. Illustrious XL natively supports 1536x1536 resolution, blending advanced Natural Language Processing with structured Danbooru tags to deliver flawless prompt adherence and stunning detail straight out of the box.
The new foundation for open-source AI illustration
Developed by OnomaAI Research, Illustrious XL solves the native resolution and prompt-adherence bottleneck, serving as the perfect untuned canvas for artists, fine-tuners, and developers.
1536px Native Resolution
Hybrid Prompting System
Stable Text Encoder
Superior Anatomy
Danbooru2023 Trained
LoRA & Finetune Ready
Break free from the 1024x1024 resolution limit
Standard SDXL models cap native generation at 1024px, forcing creators to rely on upscalers that hallucinate or blur fine lines. Illustrious XL trains natively at 1536px, unlocking unparalleled sharpness, intricate clothing details, and perfect aspect ratios.
Harmonized NLP and Booru Tags
Write a descriptive English sentence and append specific anime tags. The model interprets both seamlessly for maximum creative control.
The Ultimate Base Canvas
Deliberately left untuned for a specific aesthetic, making it an incredibly stable parent model for training custom checkpoints and LoRAs.
Why Illustrious XL dominates the anime AI ecosystem
Digital artists, model finetuners, and power users don't just want random images—they need precise character control and professional-grade clarity. Illustrious XL bridges the gap between casual prompting and professional illustration.
Extreme detail and color dynamics
Achieve unprecedented visual fidelity with highly appealing color science and anatomical accuracy.
- Flawless character anatomy and hands
- Rich, vibrant, and dynamic color palettes
- Maintains micro-details in complex outfits
Seamless UI integration
Drop the model into your existing workflow and start generating immediately with native compatibility.
- Full support for ComfyUI, A1111, and Forge
- Official VAEs and custom ComfyUI nodes provided
- Highly predictable generation behaviors
Continuous, aggressive iterations
Benefit from a model that evolves. OnomaAI has rapidly refined the model from v0.1 through v3.5, heavily optimizing stability.
- v1.0 introduced the 1536px milestone
- v2.0 STABLE brought cosine annealing for predictable rendering
- v3.5 VPred pushed SDXL to its absolute technical limits
How to create with Illustrious XL
Experience a highly responsive, state-of-the-art workflow that turns your imagination into high-resolution masterpieces.
01
1. Set up your environment
Load the Illustrious XL checkpoint into ComfyUI, Automatic1111, or Draw Things.
02
2. Craft a Hybrid Prompt
Combine natural conversational English with Danbooru2023 tags (e.g., 'A girl sitting in a cafe, 1girl, red hair, sunlight').
03
3. Set Native Resolution
Set your generation dimensions to 1536x1536 or equivalent aspect ratios for optimal clarity.
04
4. Apply LoRAs and Generate
Attach community-trained character or style LoRAs and generate your production-ready illustration.
Breakthrough features for AI power users
Packed with advanced research and optimization, Illustrious XL sets the benchmark for open-source generative models.
Stable Text Encoder Finetuning
Unlike standard practices that freeze the text encoder, OnomaAI achieved stable fine-tuning, yielding vastly superior concept understanding.
- Deep comprehension of complex actions
- Follows intricate, multi-subject prompts
- Reduces prompt bleed between characters
Massive Specialized Dataset
Trained on the comprehensive Danbooru2023 dataset, giving it an incredibly vast knowledge of character designs and artistic techniques.
- Recognizes thousands of specific anime characters
- Understands varied artistic styles (cel-shaded, painting, 3D)
- Cutoff knowledge up to June 2024
v3.0 EPS & VPred Architecture
The latest SDXL iterations utilize V-Prediction parameters for enhanced contrast, dynamic range, and overall compositional stability.
- Deeper blacks and brighter highlights
- Stronger adherence to difficult perspectives
- Faster convergence during generation
Open Source Ethos
Released under the CreativeML OpenRAIL-M and Fair AI Public License, fostering a transparent and collaborative community.
- Free for personal and research use
- Encourages community fine-tuning
- Prohibits closed-source proprietary monetization
The Bridge to Lumina
Having maximized the SDXL architecture with v3.5, the project now pioneers the newer Lumina Image architecture (Illustrious LU).
- Next-generation research in progress
- Lumina architecture for better scalability
- Early v0.03/v1.0 weights already available
Unrivaled Community Ecosystem
Serving as the base for thousands of derivatives, it powers a massive ecosystem of creator tools on platforms like Civitai.
- Thousands of compatible LoRAs available
- Active developer and researcher support
- Extensive workflow sharing
Redefining the standard for AI Illustration
Pioneered by OnomaAI Research, pushing SDXL technology to its absolute limit.
1536x1536
Native Resolution
Generates natively at ultra-high resolutions without needing external upscalers.
Hybrid NLP+Tags
Prompt Architecture
Seamlessly fuses conversational English with structured booru tagging.
v3.5 VPred
Latest Version
The definitive final SDXL iteration before transitioning to Lumina.
What creators are saying about Illustrious XL
The preferred choice for AI artists, LoRA trainers, and researchers who need absolute control and high-resolution outputs.
★★★★★
The 1536px native resolution completely changed my workflow. I no longer have to rely on Hires.fix to get sharp lineart and details.
A
AI Illustrator
Freelance Artist
★★★★★
As a LoRA trainer, Illustrious v1.0 and v2.0 provided the cleanest, most untuned canvas I've ever worked with. The concept separation is unreal.
M
Model Finetuner
Civitai Creator
★★★★★
Being able to use natural language alongside Danbooru tags gives you the exact precision of tags with the complex composition understanding of NLP.
W
Workflow Developer
ComfyUI Community
Frequently asked questions
Everything you need to know about Illustrious XL, its iterations, and how to use it.
Illustrious XL (ILXL) is an open-source, SDXL-based generative AI model created by OnomaAI Research, optimized for high-resolution 1536x1536 anime-style art and illustration.
Unlike standard SDXL (1024px), ILXL natively generates at 1536px. Compared to Pony Diffusion, ILXL acts more as an untuned, stylistically neutral 'base canvas', making it highly flexible for LoRA training and custom checkpoints without forcing a specific art style.
ILXL uses a Hybrid Prompting System. You can write a natural English sentence describing the scene, followed by a comma-separated list of Danbooru tags for specific details (e.g., character traits, clothing, background elements).
v1.0 introduced the 1536px resolution. v2.0 STABLE brought improved fine-tuning stability and predictable color science via cosine annealing. v3.5 VPred is the final SDXL iteration, utilizing V-Prediction for maximized dynamic range and structural adherence.
Illustrious XL is released under open-source licenses (CreativeML OpenRAIL-M / Fair AI Public License). However, OnomaAI explicitly prohibits using the model for closed-source proprietary monetization.
It is fully compatible with standard AI interfaces including ComfyUI, Automatic1111, Forge, and Draw Things. OnomaAI also provides custom nodes for ComfyUI for advanced users.
Illustrious LU (Lumina) is the next-generation research project from OnomaAI. Having pushed SDXL to its limits with ILXL v3.5, they are now building on the newer Lumina Image architecture for future models.
Experience high-resolution AI art natively
Stop compromising on resolution and prompt adherence. Build your next masterpiece with Illustrious XL today.