V2: expanded the amount of images in the dataset and trained it for longer, should produce more flexible and higher fidelity results. It should be noted that it wasn't trained on their nsfw artwork, so this is just to replicate their general style not their content. V1: This was a request for a LORA for the artist Sidkid; they have a cartoonish style. I think it's mostly represented here. It was trained on 50 images for around 900 steps. The trigger tag should be SidKidStyle.

SidKid Style LORA

Expanded the dataset, removed the need for trigger words, should give it more flexibility and higher quality outputs.

Files with SHA256: 0ef3c540251cdc5e93ac7a2f8ddd1de42d0de046de5d6a198eace9c58fac90e2