TheaVanherst Style Lora (Illustrious)

详情

模型描述

请先阅读“关于此版本”部分中的版本详情和建议。

LoRA,基于兽迷艺术家TheaVanherst在Civitai上训练

或称为 thea_vanherst / vanh / 𝖁𝖆𝖓𝖍_the_deer_queen / vanh.art / vanh-art

请通过以下链接支持艺术家:

https://www.patreon.com/TheaVanherst

https://www.furaffinity.net/user/theavanherst

https://vanh.art

https://www.twitter.com/TheaVanherst

https://bsky.app/profile/vanh.art

https://ko-fi.com/theavanherst

https://www.redbubble.com/people/Vanh-art

  1. 始终使用标签 by theavanherst

  2. 此LoRA基于 Illustrious 模型训练

  3. 尽管使用了Illustrious基础模型,但在Civitai中使用其他模型时,图像可能出现一些不一致,尤其在眼睛和手部。若在Civitai中使用,请搭配一些Embedding / Face Fix以获得更佳效果。对于使用ComfyUI的用户,建议搭配Hirefix / DDetailer以改善角色面部细节。

  4. 在Civitai中调整LoRA权重由您自行决定

  5. 为在Civitai中更好地使用LoRA,我建议使用VAE (SDXL VAE) 以减少手部和眼睛细节的错误。

  6. 如果您希望生成彩色图像,请将monochrome, greyscale, sketch标签加入负面提示中,因为该LoRA是基于黑白图像或单一色系训练的。

  7. 请先检查与您选择的图像风格兼容的模型,某些模型会影响LoRA在最终图像生成中的效果,例如:兽迷、人类、写实图像等。因此,同一模型不一定适用于所有类型的图像。

  8. 如果您希望强制图像生成器使用您选择的LoRA权重,请在正向提示中加入以下标签:

Illustrious (v1.0)

<lora:TheaVanherst_Style_Lora_Illustrious:1>

注意:LoRA名称末尾的数字(1)代表LoRA权重。在图像生成器中我们有滑块调节权重,但若想“强制”使用,请在正向提示中使用与生成器滑块相同的数值,例如:

Illustrious (v1.0)

Civitai滑块 = 1

正向提示: <lora:TheaVanherst_Style_Lora_Illustrious:1>
  1. 我添加了与艺术家最知名角色相关的标签。如需生成接近特定角色的图像,请务必输入角色名称。以下为每个角色的正向提示示例

Izabela Deblanc

by theavanherst, monochrome, greyscale, izabela deblanc (theavanherst), furry, furry female, anthro, animal ears, animal nose, snout, afghan hound, canine, domestic dog, sighthound, tan fur, body fur, blue eyes, long hair, blonde hair, big breasts, huge breasts, wide hips, tail, 1girl, solo, large breasts, ponytail, smile, closed mouth, half-closed eyes, looking at viewer, blush, armor, belt, cleavage, upper body, simple background

Vahn

by theavanherst, monochrome, greyscale, vahn (theavanherst), furry, furry female, anthro, animal ears, animal nose, snout, deer, brown fur, tan fur, body fur, two tone fur, (yellow sclera, black eyes:1.4), antlers, short hair, brown hair, green hair, multicolored hair, big breasts, wide hips, thick thighs, huge ass, tail, 4 ears, 1girl, solo, sweater, long sleeves, skirt, pleated skirt, sitting, on chair, office chair, crossed legs, knee up, jewelry, earrings, necklace, holding drinking glass, hand up, wine glass, looking at viewer, half-closed eyes, open mouth, boots, indoors, computer, alternate costume, feet out of frame, blush, sideways glance, table, heart symbol, keyboard \(computer\)

Bastet

by theavanherst, monochrome, greyscale, bastet (theavanherst), furry, furry female, anthro, animal ears, animal nose, snout, domestic cat, feline, egyptian mythology, egyptian god, middle eastern mythology, egypt, egyptian, purple fur, body fur, (yellow sclera, green eyes:1.4), eyelashes, makeup, long eyelashes, short hair, bob cut, dark blue hair, big breasts, huge breasts, hyper breasts, wide hips, thick thighs, dark blue tail, 1girl, solo, egyptian, egyptian clothes, navel, jewelry, full body, looking at viewer, spear, holding polearm, polearm, weapon, space, cosmic, stars, flying, black background

注意:艺术家使用了许多广为人知的兽迷角色(卡通、电影、漫画等),但在训练中我使该LoRA保持灵活,可与其他LoRA混合适配。我建议您使用特定角色的LoRA,以获得更满意的结果。

以下为初始配置,但您可自由选择其他配置:

ILLUSTRIOUS

模型

WAI-illustrious-SDXL - v14.0

LoRA

TheaVanherst Style Lora v1.0 (illustrious) - 0.5 ~ 1.0 (建议使用1)

高级设置

CFG Scale - 3.0 ~ 9.0 (建议使用7)

采样器 - DDIM / Euler a / Euler / Heun / DPM2 / DPM++ 2M Karras / Uni PC (建议使用Euler a)

步数 - 20 ~ 60 (建议使用30)

Clip Skip - 1/2

VAE - SDXL VAE

正向提示

masterpiece, best quality, amazing quality, very aesthetic, high resolution, ultra-detailed, absurdres, newest, scenery, depth of field, scenery, volumetric lighting, by theavanherst, furry, furry female, anthro, animal nose, snout, animal ears, animal nose, snout, crash bandicoot (series), tawna bandicoot, bandicoot, marsupial, green eyes, blonde hair, long hair, ponytail, orange fur, tan fur, body fur, two tone fur, makeup, lipstick, lips, 1girl, solo, big breasts, puffy nipples, huge areolae, navel, pussy, thick thighs, voluptuous, completely nude, from above, high-angle view, walking, barefoot, feet, shadow, full body, (grey background, simple background:1.4)

负向提示

bad quality, worst quality, worst detail, censor, monochrome, greyscale, sketch

一般建议:

  • 如有可能,请在使用此LoRA时注明作者TheaVanherst的署名。

  • 请勿出售或商业化此LoRA。

  • 请勿移除由该LoRA生成图像的元数据。

此模型生成的图像

未找到图像。