Katheryn Winnick, a Canadian actress.

All the sample images were generated by the flux1-dev-fp8-e4m3fn model.

So far, I've found that it's pretty good at following prompts, but the images it creates just don't have a lot of punch. Maybe I'm not using it right, though.

We need to fine-tune the F1 model more, just like we did with SD1.5, to improve its visual output. However, the cost of doing so is significantly higher, making it a more challenging task.