A LoRA trained on 22 images of PC-98 pixel art. Captures the dithering effect!

It seems to do best when prompted with just landscapes or a setting with objects. Though you could try turning down the strength a bit to get some nice results with other prompts.

The sweet spot on the strength seems to be 0.66 for most prompts! I usually run it for 80-90 steps.

This is my first public LoRA. Please let me know what you think! ❤️

I trained it using FluxGym - https://github.com/cocktailpeanut/fluxgym