[2024-11-7] Updated, added a illustrious-XL version.illustrious-XL version works best with finetuned models like T-illunai3 or Soft_Noob, control output style in illustrious-XL or NoobAI-XL base model is harder. First attempt on training all in one character LoRA.<s>For some reason on the training labeling, you have to </s><s>describe the character appearance in detail</s><s> to get the character. The less popular the character, the more description is needed. 您必须详细描述角色的外貌，才能生成角色。越是冷门的角色，就越需要描述。キャラクターを獲得するためには、キャラクターの外見を詳細に記述する必要がある。人気が高くないキャラクターほど、より多くの記述が必要になる。</s>(Version after 1.5 simplified trigger words) This LoRA included 9 characters from the game VA-11 Hall-A: Cyberpunk Bartender Action. They are: Jill Stingray, Anna Graem, Dorothy Haze, Alma Armas, Dana Zane, Stella Hoshii, Sei Asagiri, Kira Miki and Nicole Chen (a.k.a. Streaming-Chan)And an extra Model Warrior Julianne outfit for jill.For example, to get anna graem, example trigger words will be: anna graem, black serafuku, red neckerchief, purple hair, purple eyes, blunt bangs, bob cutIf you had any trouble in generating a character, please leave comments below, and I will try to help.If you generated some awesome pictures using this lora, please share them in the comments section.

[character]VA-11 Hall-A Characters (All-in-one)

Made some data labeling changes to make this LoRA easier to use.Results are way less overfitted and it now can be presented with more styles. You can recall characters with fewer keywords than v1. However, details like Sei Asagiri's armor are not as accurate as v1.

Files with SHA256: 083305cc1563ac93c2030643e26ea3acb6871f4ad621994c3c427033e3fabaa7