Usnarbit Style Lora (Illustrious / Pony)
详情
下载文件
关于此版本
模型描述
请先阅读“关于此版本”部分中的版本详情和建议。
在 Civitai 上基于毛茸茸艺术家 Usnarbit 训练的 LoRA
请通过以下链接支持该艺术家:
https://www.furaffinity.net/user/usnarbit/
https://bsky.app/profile/usnarbit.bsky.social
https://www.patreon.com/usanrbit
始终使用标签 by usnarbit
此 LoRA 基于 Pony V6 版本 / Illustrious
尽管使用了 Pony V6 基础模型,但在 Civitai 上使用其他模型时,图像中某些部分(特别是 眼睛和手部)可能出现不一致。若在 Civitai 上使用,请搭配某些 Embedding / Face Fix 以获得更好效果。对于使用 ComfyUI 的用户,建议使用 Hirefix / DDetailer 来改善角色面部图像质量。
别忘了使用与 Pony 模型相关的标签:
score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,
在 Civitai 上调整 LoRA 权重由您自行决定
为了在 Civitai 上更佳地使用 LoRA,我建议使用 VAE (SDXL VAE) 和 Embedding (Pony PDXL Negative Embeddings),以减少手部和眼睛细节中的错误
请先检查与您所选图像兼容的模型,某些模型会影响 LoRA 在最终图像生成中的效果,尤其对以下类型如:毛茸茸、人类、写实图像等。因此,同一模型未必总适用于不同类型的图像。
如果您希望强制图像生成器使用您选择的 LoRA 权重,请在正向提示词中使用以下标签:
Illustrious (v1.0)
<lora:Usnarbit_Style_Lora_Illustrious:1>
Pony (v1.0)
<lora:Usnarbit_Style_Lora:0.8>
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,但如果您想“强制”使用特定值,请在正向提示中使用与图像生成器滑块相同的数值,例如:
Illustrious (v1.0)
Civitai 滑块 = 1
正向提示: <lora:Usnarbit_Style_Lora_Illustrious:1>
Pony (v1.0)
Civitai 滑块 = 0.8
正向提示: <lora:Usnarbit_Style_Lora:0.8>
注意:该艺术家使用了多个知名毛茸茸角色(卡通、电影、漫画等),但在训练中我使 LoRA 保持灵活,可与其他 LoRA 混合使用。我的建议是使用特定角色的 LoRA,以获得更满意的最终效果。
以下为初始配置,但您可自由选择其他配置:
ILLUSTRIOUS
模型
Nova Furry XL - Illustrious v6B
LoRA
Usnarbit Style Lora - v1.0 (illustrious) - 0.5 ~ 1.0 (推荐 1)
高级设置
CFG Scale - 3.0 ~ 9.0 (推荐 6)
采样器 - DDIM / Euler a / Euler / Heun / DPM2 / DPM++ 2M Karras / Uni PC (推荐 Euler a)
步数 - 20 ~ 60 (推荐 30)
Clip Skip - 1/2
VAE - SDXL VAE
正向提示词
masterpiece, best quality, amazing quality, very aesthetic, high resolution, ultra-detailed, absurdres, newest, scenery, depth of field, scenery, volumetric lighting, by usnarbit, furry, furry female, anthro, toriel, undertale, convenience store, shelves, groceries, grocery store, shopping kart, 1girl, solo, animal ears, from behind, tail, pants, purple shirt, thick thighs, horns, goat ears, huge breasts, huge ass, goat girl, standing, short sleeves, indoors, black pants, snout, looking object, holding object, holding cereal box, plump, hand up, body fur, mature female, t-shirt, goat horns, white fur, curvy, voluptuous, curvaceous, tight bottomwear, tight clothing, tight pants, high-angle view
负向提示词
bad quality, worst quality, worst detail, bad quality, worst quality, worst detail, sketch, censor, human, multiple tails, modern, recent, old, oldest, graphic, cartoon, text, painting, crayon, graphite, abstract, glitch, deformed, mutated, ugly, disfigured, long body, lowres, bad anatomy, bad hands, missing fingers, extra digit, fewer digits, cropped, very displeasing, (worst quality, bad quality:1.2), bad anatomy, sketch, jpeg artifacts, signature, watermark, username, simple background, conjoined, ai-generated, worst quality, low quality, off-topic, old, early, chromatic aberration, abstract, jpeg artifacts, scan artifacts, logo
PONY
模型
WAI-ANI-NSFW-PONYXL - v13.0
LoRA
Usnarbit Style Lora - v1.0 (Pony) - 0.5 ~ 1.0 (推荐 0.8)
高级设置
CFG Scale - 3.0 ~ 9.0 (推荐 6)
采样器 - DDIM / Euler a / Euler / Heun / DPM2 / DPM++ 2M Karras / Uni PC (推荐 DPM++ 2M Karras)
步数 - 20 ~ 50 (推荐 30)
Clip Skip - 1/2
VAE - SDXL VAE
正向提示词
score_9, score_8_up, score_7_up, by usnarbit, furry, furry female, anthro, lagomorph, leporid, rabbit, 1girl, solo, mature female, milf, underwear, medium breasts, animal ears, floppy ears, bunny ears, blonde hair, long hair, panties, wide hips, black panties, makeup, navel, off shoulder, thick thighs, shirt, eyeshadow, standing, full body, off-shoulder shirt, smile, pink shirt, crop top, black eyes, barefoot, looking at viewer, orange fur, body fur, animal nose, bare shoulders, :3, plump, long sleeves, buck teeth, lipstick, stomach, highleg panties, hand up, curvy, indoors, midriff, highleg, two-tone fur, collarbone, topknot, pantsless
负向提示词
zPDXL3, 3D, 2D, human, blurry, jpeg artifacts, username, watermark, signature, normal quality, worst quality, large head, low quality, text, error, missing fingers, extra digits, fewer digits, bad eye
由于此模型基于 Pony Diffusion,负向提示词中无需过多指定,因为 Pony 模型本身非常灵活,但我始终保留 zPDXL-neg 标签(/model/332646/pony-pdxl-negative-embeddings)。
通用建议:
使用此 LoRA 时,请尽可能注明作者 Usnarbit 的署名。
请勿销售或商业化此 LoRA;如有可能,请通过您的 Patreon 帮助该作者。
请勿移除由 LoRA 生成图像的元数据。











