Although this is the first version of ArtiWaifu Diffusion, it serves as a powerful successor to the Anime Illust Diffusion XL (AIDXL) series of models. Rather than marking the conclusion of the AIDXL series, it represents a significant step forward, advancing the series to become even more robust and capable.
🏷 Model Information 🏷
Developed by: Euge
Funded by: Neta.art
Model type: Generative text-to-image model
Finetuned from model: SDXL 1.0 Base
License: Fair AI Public License 1.0-SD
Support a lot of painting styles, art styles (9000+) and characters (6000+)
Can accurately identify specific quality level, safety level, age of drawing, vintage of drawing and some aesthetic concepts.
🚀 Dataset Source 🚀
Subset of Danbooru (1.3M). Selected by aesthetic score using Waifu Scorer V3.
Subset of Pixiv (0.1M). Hand-selected.
Others (0.1M).
👀 Future Plan 👀
Apply stronger training strategies.
Continue adding more and more high-quality data.