🌟 Meet Rick LoRA - The Ultimate April Fools' Overfit Experiment 🌟

This LoRA was intentionally overfit on just one image (watch the animated version of it here). Training data was 30 copies of this image with random prompts, causing the LoRA to ignore prompts entirely—as you can verify on the Civitai example generations. Works best for standard tall images; on other aspect ratios, the text becomes unreadable.

Consider this a proof of concept or historical record—you never really know what's inside a .safetensors file. In this case, 243 MB is just one image. The most inefficient image compression ever.

This model was originally published on April 1st under the guise of a "3D MMO Unreal Style for ZIT" LoRA, complete with example images of 3D waifus as a facade. The actual LoRA used to generate those example images can be found here. A huge thank you to everyone who downloaded, liked, and played along with the act on April 1st.

Disclaimer: This model is provided "as is," which is to say, "as Rick." Manage your expectations accordingly. Or don't. He's never gonna give you up, anyway.