Updated: Oct 11, 2026
style
AniFusion XL v2 - a refined difference merge of Illustrious-XL-v1.0 and NoobAI-XL-v1.1 on the SDXL base.
Built as: Illustrious-XL-v1.0 + 0.20 x (NoobAI-XL-v1.1 - SDXL-base-1.0), applied to the UNet only (VAE and text encoder kept from Illustrious).
Why NoobAI: it is itself a fine-tune of Illustrious, so this difference merge only carries over its extra line clarity and micro-detail - it does not drag in a different art style. Compared with v1 you should see sharper linework and richer detail while keeping Illustrious' composition and prompt adherence.
[Recommended] Sampler DPM++ 2M / DDIM, 28-32 steps, CFG 7-7.5, 832x1216. Use bf16 where possible - fp16 can soften the output noticeably.
[License] CreativeML Open RAIL-M (inherited from the base models).