This character is from Etuzan Jakusui. The LoRA was trained using a small set of raw sketches, and I manually refined some images and even drew certain details like glasses, sweat effects, and wet textures to improve consistency.
The results weren't perfect in those areas, but overall, this version works better than my previous one.
Training and testing took over 10 hours — thanks to my potato PC! 😒😒😒
For best results, use detailed, descriptive prompts. Avoid simple prompts, as they tend to generate random or poor-quality images.
Description
This character is from Etuzan Jakusui. The LoRA was trained using a small set of raw sketch images.
To achieve the best results, use a detailed and specific prompt. Avoid simple or vague prompts, as they may generate random or low-quality outputs