<h1 id="heading-10">LoRA</h1>A LoRA for Countess from Paragon. Please use an upscaler for better results! I recommend using R-ESRGAN 4x+.All sample images are generated with my custom merged model <a target="_blank" rel="ugc" href="https://civarchive.com/models/23860/umbramix">UmbraMix</a>I've uploaded various checkpoints from the training, choose the one which fits the most your needs!The soft versions are versions that retained some detail about countess such as the hair cut, clothing style, but the charater may not be recognizable at first.The hard versions are versions where the LoRA has hard backed details about Countess which you may not be able to remove since they're hard backed onto the learning process. <h2 id="heading-11">Prompts</h2><ul><li>Other relevant tags: vampire, red hair, black hair, latex, bob cut, 1girls, solo</li><li>All sample prompts are generate with a CLIP SKIP value of 2</li></ul><h2 id="heading-12">Weights</h2>I recommend using the fallowing weights:<ul><li>Closest to the character : 0.75 ~ 1.0</li><li>When trying to add / change the character clothing/assets : 0.35~0.7</li></ul>

Countess (paragon)

Details hardly backed onto the LoRA, removing details about Countess may need a very low weight which can cause huge detail loss

Files with SHA256: 3dc8e2aad76e2b2caf42bd474e112ed61e4829885dac8b494faac4ca434958bb