This model is a merge of NoobAI-XL (NAI-XL) - V-Pred 1.0 and RouWei v0.8, plus a little spice of my Style LoRA to add a certain aesthetic. All configured in V-Prediction.
I created this merge with artist styles in mind, so it can recognize any artist name from Danbooru up to April 2025, thanks to the RouWei dataset. However, even if you don't want to use artist styles, this model is already good enough to produce high-quality images.
Recommended Settings
Positive prompt: masterpiece, best quality, very awa, very aesthetic, highres, absurdres
Negative prompt: worst quality, low quality, bad quality, watermark, worst aesthetic,
CFG: 3 ~ 5
Step: 20 ~ 30
Sampler: Euler a
Recommended Resolution
768x1344 (9:16)
832x1216 (2:3)
896x1152 (3:4)
1024x1024 (1:1)
1152x896 (4:3)
1216x832 (3:2)
1344x768 (16:9)
Use ADetailer and Upscale to get the best result.
Hires upscale: 1.5
Hires steps: 20,
Hires upscaler: R-ESRGAN 4x+ Anime6B,
Denoising strength: 0.3 ~ 0.5