"They worshiped, so they said, the Great Old Ones who lived ages before there were any men, and who came to the young world out of the sky. Those Old Ones were gone now, inside the earth and under the sea; but their dead bodies had told their secrets in dreams to the first man, who formed a cult which had never died. This was that cult, and the prisoners said it had always existed and always would exist, hidden in distant wastes and dark places all over the world until the time when the great priest Cthulhu, from his dark house in the mighty city of R'lyeh under the waters, should rise and bring the earth again beneath his sway."― H.P. Lovecraft, <a target="_blank" rel="ugc" href="https://www.goodreads.com/work/quotes/25692046">The Call of Cthulhu</a> Update:I was having so much fun that I decided to train the next version immediately. This time on Clip Skip 2, so it's compatible with all your favorite anime models (I will be training the beta 4 version on Clip Skip 1 later as well). Due to the larger dataset, you're going to see more consistency, although the form is overall going to be more "human-like" in appearance and body type. You can keep the weight a bit higher as well.Original Post:Hello folks, This LoRA was borne out of training my Delta Green style LoRAs and TI. While SD has a faint idea of what Cthulhu looks like, I thought that I could help it out. Now, you don't have to choose the lesser evil, and can summon the Great Cthulhu in all his terrible glory. Usage Notes:<ul><li>The model is trained on multiple interpretations of Cthulhu, so while some features are a through-line expect a fair degree of variability regardless of weight.</li><li>Trained on Clip Skip 1.</li><li>Weight should be fine from 0.6 to 0.9.</li></ul>

The Great Cthulhu - Character LoRA

Upped the dataset size to 40 images and trained on animefull.

Files with SHA256: 3d89d13e6f58a9e5da98d67f4595f8df3cb8ff62db0eaf7d4d07d920de5f61a0