Reton v2
Custom SDXL LoRA trained at 1024px for 3000 steps. Trigger word: retonstyle.
Recommended: 30 inference steps, guidance scale 5.0, 1024x1024. Negative prompt: blurry, low quality, watermark, text artifacts, distorted, amateur, cluttered.
LoRA weights use PEFT-style keys (unet.base_model.model.*). Four generated sample images are in samples/.
Note: SDXL can create attractive designs but exact typography and spelling remain less reliable than purpose-built modern image generation models. For production graphics, overlay final text with a typography/layout tool.
- Downloads last month
- 27
Model tree for Erog0291/reton-v2
Base model
stabilityai/stable-diffusion-xl-base-1.0