It takes substantial time and efforts to bake models. If you appreciate my models, I would be grateful if you could support me on Ko-fi ☕ .
🚀 Please also check out Selai Mix, my recent checkpoint merge model release, if you love the aesthetics of Husbando LoCon!
Easy to create good-looking masculine characters using tag-based prompts.
Good male genitalia.
Good solo / solo focus homoerotic content.
Bara/yaoi ?????? depicting 2 or more people are highly erroneous. You'll need luck and careful prompting to get things right.
CLIP skip: 2
You can start by trying a weight of 0.5 with CFG scale 7.
For ?????? art, I recommend LoRA weight of 1.0 with CFG scale 9.
Since Selai Mix already contains Husbando LoCon v4.0, a lighter weight such as 0.3 should already give you the desired effect. You should not set the weight to over 0.7 when using on Selai Mix.
Select any checkpoint that works with tag-based prompts. Recommended models: Selai Mix and AnyLoRA.
Select a VAE if the checkpoint model you're using doesn't have a baked-in VAE, or else you'll find your generated images have dull colors. Recommended VAEs: blessed2, vae-ft-mse-840000-ema-pruned or kl-f8-anime2.
Make sure to follow the steps in How to use a LoRA.
Generate at 512 base resolution (512x768, 512x512 or 768x512) and use hires fix to improve the details. Recommended upscaler: R-ESRGAN 4x+ Anime6B; denoising strength: 0.4.
Fixed the overfitting problem in previous versions.
Even larger training dataset (6,617 images)
Auto tagged with wd-v1-4-moat-tagger-v2 with zero pruning.
Used flip augmentation and color augmentation.
Used multi resolution noise during training.
Obtained from LoCon extraction after fine-tuning the entire model.
Downscaled the LoCon dimension to 128.
1. I diritti dei modelli ripubblicati appartengono ai creatori originali.
2. I creatori originali che desiderano reclamare il proprio modello devono contattare il personale di SeaArt AI tramite canali ufficiali. Clicca per rivendicare
