Sometimes when creating storyboards it's helpful to actually have something that's deliberately low fidelity. When you have something that's too high-fidelity, sometimes people get hung up on details in your storyboard that are inconsequential to what you're hoping to communicate.

The intent of this SDXL LoRA is to give you the ability to hopefully create low-fidelity sketches with enough consistency from image to image so your final product is coherent.

The LoRA was trained on 100 sketch images that were originally created in Midjourney using easily identifiable subjects and settings. These training images were then color corrected to minimize color variation in the output.

I hope you find useful, and I welcome any feedback.