Related to Ink Art models I trained for Flux and SDXL, but based on an altogether different dataset - this model achieves much better what I was attempting in previous versions.

I preface prompts simply with "Illustration of ... " and usually that's all I need. Some images may turn toward photography or more cartoony than I like, and so i may reinforce the aesthetic by ending the prompt with "graphic novel style illustration".

The LoRA is trained on Z-Image Base, but my images are all generated with Z-Image Turbo.