Тип
LORA
Время публикации
2023-10-11
Базовая модель
SD 1.5
Версия введения.
Current version will
1) only work with realism models (or anything trained with SD 1.5 as the base model)
2) will likely need hires fix or other kinds of upscaling to produce good results. If images look fried then try setting the CFG Scale to a low-ish number like 4 or 5. If you want or need the CFG scale set to a higher number (this can help with nudging the model to more strictly follow your prompt) then i suggest using HiRes fix set to 0.3 with the lollypop upscaler.
