I've created a new model on Stable Diffusion 1.5 for generating vampire portraits! Using a variety of sources such as movies, novels, video games, and cosplay photos, I've trained the model to produce images with all the classic vampire features like fangs and glowing eyes. While some images may require a bit of cleanup or more detailed prompts, you can expect to get great results with just basic prompts using the trigger word "OOKZA Vampire Style", eg. "OOKZA Vampire Style portrait of (Sarah Michelle Gellar) as a Vampire, dramatic lighting, intricate, gothic, highly detailed, digital painting, trending on artstation, concept art, smooth, sharp focus, illustration, 8k, photo realistic, masterpiece." This model also works quite well with IMG2IMG to suppliment your existing vampire portraits with gaping, hissing vampire mouths.

Vampire Style

This is a new version of the Vampire style model from scratch using a different Dreambooth colab. The results are improved from version 1.0 and in this version the Trigger words are OOKZAVampire Style (two words and not three as in the previous version). Certain vampire themed terms such as fangs, canine teeth, sharp teeth, hissing, screaming, snarling, smiling, baring teeth, exposed teeth will improve or change the output depending on your requirements.I've found that keeping the CFG scale between 7-12 generally gives the best results for good looking fangs with higher CFGs seemingly creating an "oversharpened" look on them, but higher CFG scale will also start drawing additional horror themed elements such as blood drips and splashes. One might composite different versions of the render to get the best results for good looking teeth combined with blood drips and splashes.EULER A with 20-40 steps and DPM++2S a Karras, and DPM++ 2M Karras with 30+ steps seem to provide decent image quality.

Files with SHA256: e40b980e55f4dcb6c8fe41b6bc6651c2bc1294bdf22d537d10a07e4010dc8b90