DownloadNoobAI XL 1.1 is a diffusion-based text-to-image generative model developed by Laxhar Dream Lab, fine-tuned from NoobAI XL 1.0 and built upon Illustrious-xl. It leverages the latest Danbooru and e621 datasets, using native tag captions to enhance visual fidelity, style accuracy, and prompt responsiveness. The model introduces refined quality tagging, ranking images by percentile to ensure results reflect modern aesthetic preferences. It supports a range of recommended resolutions around 1024×1024, maintaining high consistency across varied aspect ratios. NoobAI XL 1.1 prioritizes safe and ethical content generation, prohibiting NSFW and harmful outputs under the Fair-AI Public License 1.0-SD, which also forbids commercial use. The model encourages open-source collaboration, requiring derivative models and LoRAs to be shared under the same terms. Sponsored by Lanyun Cloud, it represents a community-driven effort to democratize advanced, high-quality visual generation.
Features
- Diffusion-based text-to-image model fine-tuned from NoobAI XL 1.0
- Trained on full Danbooru and e621 datasets with native tag captions
- Optimized for resolutions near 1024×1024 with multiple aspect options
- Implements percentile-based quality tagging (e.g., masterpiece, best quality)
- Enforces ethical use under Fair-AI Public License 1.0-SD (non-commercial)
- Requires open-source sharing of derivatives, LoRAs, and merged models
- Includes dedicated ControlNet models (normal, depth, and canny)
- Recommended sampling settings: Euler a, 25–30 steps, CFG scale 5–6
Project Samples
