AI绘画模型和插件汇总

绘画模型(*.ckpt/*.safetensors)

Novel AI 泄露版模型

  • 基于Stable Diffusion网络结构,使用Danbooru数据集进行多次训练后的权重
  • 至少需 5.8GiB 存储空间
  • 磁力链接,使用BitTorrent协议下载
1
magnet:?xt=urn:btih:5bde442da86265b670a3e5ea3163afad2c6f8ecc
  • SFW(Suitable For Work)模型:stableckpt/animesfw-final-pruned
  • NSFW模型:stableckpt/animefull-final-pruned

Stable Diffusion 官方模型

1
wget https://huggingface.co/runwayml/stable-diffusion-v1-5/resolve/main/v1-5-pruned-emaonly.ckpt
1
wget https://huggingface.co/stabilityai/stable-diffusion-2-1/resolve/main/v2-1_768-ema-pruned.ckpt

Waifu Diffusion

1
wget https://huggingface.co/hakurei/waifu-diffusion-v1-4/resolve/main/wd-1-4-anime_e2.ckpt

Kenshi

  • 基于Stable Diffusion网络结构,强调“照片感”。是对多个模型的加权融合,包括BoChen模型、WLOP模型等;
  • Kenshi01
1
wget https://huggingface.co/SweetLuna/Kenshi/resolve/main/KENSHI%2001/KENSHI01_Pruned.ckpt

AbyssOrangeMix3

  • 基于Stable Diffusion网络结构,是对多个模型的加权融合,侧重“高质量”和“画面内的纹理效果”
  • 分为A1(平涂)、A2(厚涂;油画风)、A3(融合)三个版本
  • 至少需 2.0GB 存储空间
  • WarriorMama777/OrangeMixs · Hugging Face
1
wget https://huggingface.co/WarriorMama777/OrangeMixs/resolve/main/Models/AbyssOrangeMix3/AOM3A3.safetensors

NeverEnding Dream

1
wget https://civitai.com/api/download/models/11925

ChilloutMix

1
wget https://civitai.com/api/download/models/11732

momoko-e

  • 基于stable-diffusion网络结构,训练了ももこ画师的图片,该模型比较火的原因可能是手和脚的崩坏次数少一些(被一些人称为“有手修复包”,很可能是因为模型过拟合)
  • 至少需 2.0GB 存储空间
  • 链接来自其他人转存,有失效风险
1
wget https://huggingface.co/LarryAIDraw/momoko-e/resolve/main/momoko-e.ckpt

Anything v4.5

  • 基于Stable Diffusion网络结构,是对多个模型的加权融合及补充训练,侧重“全能(二次元男角色、二次元女角色、二次元风景)”和“更短的tag”
  • 至少需 2.0GB 存储空间
  • andite/anything-v4.0 · Hugging Face
1
wget https://huggingface.co/andite/anything-v4.0/resolve/main/anything-v4.5-pruned-fp16.ckpt

Realistic Vision V2.0

1
wget https://civitai.com/api/download/models/29460

WebUI 插件

a1111-sd-webui-tagcomplete

sd-webui-supermerger

sd-webui-controlnet

openpose-editor

Auto-Photoshop-StableDiffusion-Plugin

stable-diffusion-webui-localization-zh_CN

sd-webui-additional-networks

deforum-for-automatic1111-webui

sd-webui-depth-lib

sd-webui-cutoff

stable-diffusion-webui-promptgen

sd-webui-regional-prompter

ebsynth_utility

stable-diffusion-webui-two-shot

stable-diffusion-webui-composable-lora

stable-diffusion-webui-wd14-tagger

Textual Inversion模型

  • 引入新的prompt,并将其与已有的prompt关联起来;
  • 放置在WebUI的embeddings文件夹下;

EasyNegative

1
wget https://huggingface.co/datasets/gsdf/EasyNegative/resolve/main/EasyNegative.pt

Ulzzang-6500

1
wget https://civitai.com/api/download/models/10107

LoRA模型

  • LoRA用于训练差分模型。用户给出一系列具有相似特征(比如都是某个动漫角色、都是某个绘画风格、都是某个地区的人的脸型,……)的图片,经过训练生成一个几十/几百MB的模型
  • 载入绘画模型,再载入LoRA模型,使用LoRA的Prompt(或是插件),便可生成具有训练数据特征的图片
  • 特点:速度快,模型小(类似的Dreambooth技术会生成完整模型),准确度高(对“风格”来说更优秀,但对“脸型”来说不如Dreambooth),可组合
  • 推荐配合sd-webui-additional-networks插件使用
    • 使用插件:LoRA模型放到extensions/sd-webui-additional-networks/models/lora
    • 不使用插件:LoRA模型放到models/Lora

*-doll-likeness

  • 配合ChilloutMix绘画模型,生成具有特定地域脸型的人像图片(“伪Cosplay照片”)
  • 原作者已删除模型,链接来自他人转存,有失效风险
  • Prompt:woman, girl
1
2
3
4
5
6
# 日本模特脸型
wget https://huggingface.co/AnonPerson/ChilloutMix/resolve/main/Japanese-doll-likeness.safetensors
# 韩国模特脸型
wget https://huggingface.co/AnonPerson/ChilloutMix/resolve/main/Korean-doll-likeness.safetensors
# 中国台湾模特脸型
wget https://huggingface.co/AnonPerson/ChilloutMix/resolve/main/Taiwan-doll-likeness.safetensors

chinese-doll-likeness

1
wget https://civitai.com/api/download/models/11195

Pastel-Mix

1
wget https://civitai.com/api/download/models/7397

Glorious (Azur Lane) Spring 光荣 凉夜春雪

1
wget https://civitai.com/api/download/models/8855

Gacha splash LORA 3.1

1
wget https://civitai.com/api/download/models/26536

Anime Lineart (线稿/線画) Style 3.0

1
wget https://civitai.com/api/download/models/28907

前景龟背竹V11(foreground plant LORA)

1
wget https://civitai.com/api/download/models/15004

工具模型

  • 与绘画不直接相关,主要用于图片后处理、图片内容识别,以及图片放大

变分自编码器(VAE)

  • 改变生成结果的概率分布,表现为图片的对比度、饱和度等
  • 部分模型内置VAE,看到_bakedVAE字样可忽略
  • 最常用的VAE:NovelAI的VAE,下载:
    • 通过磁力链接,见上述NovelAI模型章节:stableckpt/animevae.pt
    • Huggingface上其他项目的转存
  • 放置在WebUI的models/VAE文件夹下
1
wget https://huggingface.co/WarriorMama777/OrangeMixs/resolve/main/VAEs/orangemix.vae.pt
1
wget https://huggingface.co/stabilityai/sd-vae-ft-mse-original/resolve/main/vae-ft-mse-840000-ema-pruned.safetensors

torch_deepdanbooru

1
wget https://github.com/AUTOMATIC1111/TorchDeepDanbooru/releases/download/v1/model-resnet_custom_v3.pt

RealESRGAN(R-ESRGAN4x)

1
wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.1/RealESRNet_x4plus.pth

RealESRGAN+Anime6B(R-ESRGAN4x + Anime6B)

1
wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.2.4/RealESRGAN_x4plus_anime_6B.pth

GFPGAN、CodeFormer

1
2
3
wget https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/GFPGANv1.4.pth
wget https://github.com/xinntao/facexlib/releases/download/v0.1.0/detection_Resnet50_Final.pth
wget https://github.com/xinntao/facexlib/releases/download/v0.2.2/parsing_parsenet.pth
1
wget https://github.com/sczhou/CodeFormer/releases/download/v0.1.0/codeformer.pth

4xUltraSharp

  • 用于二次元图片或真实图片的放大和重建。据部分测试结果,4xUltraSharp放大半厚涂或厚涂的图片时,背景和边界的效果好于RealESRGAN
  • Model Database - Upscale Wiki

下载地址:MEGA

ControlNet

  • 用于部分控制Diffusion网络的参数,例如
    • 固定生成图片的人物动作(OpenPose)
    • 通过精细线稿生成图片(Canny)
    • 通过简笔画生成图片(Scribble)
    • 通过景深生成图片(Depth)
  • lllyasviel/ControlNet: Let us control diffusion models!
  • 需要配合插件sd-webui-controlnet使用
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
# ControlNet差分模型,放置在WebUI的extensions/sd-webui-controlnet/models文件夹下
# Canny
wget https://huggingface.co/webui/ControlNet-modules-safetensors/resolve/main/control_canny-fp16.safetensors
# OpenPose
wget https://huggingface.co/webui/ControlNet-modules-safetensors/resolve/main/control_openpose-fp16.safetensors
# Scribble
wget https://huggingface.co/webui/ControlNet-modules-safetensors/resolve/main/control_scribble-fp16.safetensors
# Depth
wget https://huggingface.co/webui/ControlNet-modules-safetensors/resolve/main/control_depth-fp16.safetensors
# HED
wget https://huggingface.co/webui/ControlNet-modules-safetensors/resolve/main/control_hed-fp16.safetensors
# MLSD
wget https://huggingface.co/webui/ControlNet-modules-safetensors/resolve/main/control_mlsd-fp16.safetensors
# Normal
wget https://huggingface.co/webui/ControlNet-modules-safetensors/resolve/main/control_normal-fp16.safetensors
# Segmentation
wget https://huggingface.co/webui/ControlNet-modules-safetensors/resolve/main/control_seg-fp16.safetensors

# OpenPose需要独立的数据模型,放置在WebUI的models/openpose文件夹下
wget https://huggingface.co/lllyasviel/ControlNet/resolve/main/annotator/ckpts/body_pose_model.pth
wget https://huggingface.co/lllyasviel/ControlNet/resolve/main/annotator/ckpts/hand_pose_model.pth

第三方ControlNet

1
2
# Face Landmark
wget https://huggingface.co/OedoSoldier/Face-Landmark-ControlNet-diff/resolve/main/diff_control_sd15_landmarks_fp16.safetensors
评论