SDXL v2.0NowThe results weren't great, but made a great base to tweak and finetune though. After tweaking and finetuning the results came out good. SDXL and SD 1.5 v1 OriginallyPixel Art Lora made with 127 images synthetically generated with text to image then running them through Civit AI lora trainer

GG_Pixel Heroes

Made the base Lora on Civit, fine tuned it locally to improve output quality.

Files with SHA256: c60a083800f7712a231ebc2940b2d97616cc699c708f163f7ff90df607a39301