This LoRA was trained to produce flat-design images. The idea is to use as few lines as possible while keeping the visual clear, relying on solid blocks of color.
While the base model can already generate this style, I made an artistic choice that sets it apart from standard flat design. This choice also introduced a few biases, which may be corrected in the future if I feel like it.
For prompting, in addition to the trigger word, I suggest mentioning the colors and specifying a 2D, flat look to help avoid overly realistic and detailed generations. I hope you have fun with this model—it should be interesting to blend with some of your other favorite LoRAs. The example images were deliberatel…