The LoRA is trained using the BF16 version of Z-image base. The dataset consist of 60 Niji images given to me by my friend Stan Katayama (creator of SDXL Niji, one of my favorite models): https://civitai.com/models/120765/sdxlnijispecial-edition
Trigger: stankatayamaniji5z illustration
Repeat: 20 Epoch: 10
LR: 1.0 Scheduler: constant, Optimizer: Prodigy
Dim 32, Alpha 16





