人工智能/Stable Diffusion 安装和简单使用

Stable Diffusion 是一种通过文字描述创造出图像的 AI 模型. 它是一个开源软件, 有许多人愿意在网络上免费分享他们的计算资源, 使得新手可以在线尝试.

安装

本地部署的 Stable Diffusion 有更高的可玩性, 例如允许您替换模型文件, 细致的调整参数, 以及突破线上服务的道德伦理检查等. 鉴于我目前没有可供霍霍的 GPU, 因此我将在一台 2 核 4G 内存的云服务上部署它. 这着实非常惊人!

在安装运行 Stable Diffusion 之前, 首先需要为我的 Linux 机器创建一个 16G 大小的交换分区. Stable Diffusion 在运行过程中大概需要吃掉 12G 内存, 交换分区可以勉强让我们达到其最低运行需求. 当然如果您的机器拥有足够的内存, 可以忽略这一步.

$ dd if=/dev/zero of=/mnt/swap bs=64M count=256
$ chmod 0600 /mnt/swap
$ mkswap /mnt/swap
$ swapon /mnt/swap

安装 Python 3.10.6. 比此更高版本的 Python 暂时不支持运行 torch. 采用源码方式安装步骤如下:

set -ex

version=3.10.6

wget https://www.python.org/ftp/python/${version}/Python-${version}.tgz
tar -zxvf Python-${version}.tgz
cd Python-${version}
./configure --prefix ~/app/python-${version}
make
make install
cd ..

rm -rf Python-${version}
rm -rf Python-${version}.tgz

下载, 安装并运行 Stable Diffusion WebUI:

$ git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
$ cd stable-diffusion-webui
$ git checkout 5ef669de

$ python_cmd=~/app/python-3.10.6/bin/python3 bash webui.sh --skip-torch-cuda-test --use-cpu all --lowram --no-half --listen

等待一段时间, 在浏览器中打开 127.0.0.1:7860 即可见到 UI 界面.

下载更多模型

模型, 有时称为检查点文件, 是预先训练的 Stable Diffusion 权重, 用于生成一般或特定的图像类型. 模型可以生成的图像取决于用于训练它们的数据. 如果训练数据中没有猫, 模型将无法产生猫的形象. 同样, 如果您仅使用猫图像训练模型, 则只会产生猫.

此处介绍了一些常见的模型(v1.4, v1.5, F222, Anything V3, Open Journey v4).

Stable Diffusion WebUI 运行时会自动下载 Stable Diffusion v1.5 模型. 下面提供了一些快速下载其它模型的命令.

$ cd models/Stable-diffusion

# Stable diffusion v1.4
$ wget https://huggingface.co/CompVis/stable-diffusion-v-1-4-original/resolve/main/sd-v1-4.ckpt
# Stable diffusion v1.5
$ wget https://huggingface.co/runwayml/stable-diffusion-v1-5/resolve/main/v1-5-pruned-emaonly.ckpt
# F222
$ wget https://huggingface.co/acheong08/f222/resolve/main/f222.ckpt
# Anything V3
$ wget https://huggingface.co/Linaqruf/anything-v3.0/resolve/main/anything-v3-fp16-pruned.safetensors
# Open Journey
$ wget https://huggingface.co/prompthero/openjourney/resolve/main/mdjrny-v4.ckpt
# DreamShaper
$ wget https://civitai.com/api/download/models/5636 -O dreamshaper_331BakedVae.safetensors
# ChilloutMix
$ wget https://civitai.com/api/download/models/11745 -O chilloutmix_NiPrunedFp32Fix.safetensors
# Robo Diffusion
$ wget https://huggingface.co/nousr/robo-diffusion/resolve/main/models/robo-diffusion-v1.ckpt
# Mo-di-diffusion
$ wget https://huggingface.co/nitrosocke/mo-di-diffusion/resolve/main/moDi-v1-pruned.ckpt
# Inkpunk Diffusion
$ wget https://huggingface.co/Envvi/Inkpunk-Diffusion/resolve/main/Inkpunk-Diffusion-v2.ckpt

修改配置文件

ui-config.json 内包含众多的设置项, 可按照个人的习惯修改部分默认值. 例如我的配置部分如下:

{
    "txt2img/Batch size/value": 4,
    "txt2img/Width/value": 480,
    "txt2img/Height/value": 270
}

示例

 model: anything-v3-fp16-pruned.safetensors
prompt: colorful reflective fabric inner, pixiv, hyper detailed, futuristic fashion, anime girl, nude

img

 model: chilloutmix_NiPrunedFp32Fix.safetensors
prompt: beautiful, masterpiece, best quality, extremely detailed face, perfect lighting, (1girl, solo, 1boy, nemonelly, slight penetration, lying, on back, spread legs:1.5), street, crowd, ((skinny)), ((puffy eyes)), brown hair, medium hair, cowboy shot, medium breasts, swept bangs, walking, outdoors, sunshine, light_rays, fantasy, rococo, hair_flower, low tied hair, smile, half-closed eyes, dating, (nude), nsfw, (heavy breathing:1.5), tears, crying, blush, wet, sweat, <lora:koreanDollLikeness_v15:0.4>, <lora:povImminentPenetration_ipv1:0>, <lora:breastinclassBetter_v14:0.1>

prompt: paintings, sketches, (worst quality:2), (low quality:2), (normal quality:2), lowres, normal quality, ((monochrome)), ((grayscale)), skin spots, acnes, skin blemishes, age spot, (ugly:1.331), (duplicate:1.331), (morbid:1.21), (mutilated:1.21), (tranny:1.331), mutated hands, (poorly drawn hands:1.331), blurry, (bad anatomy:1.21), (bad proportions:1.331), extra limbs, (disfigured:1.331), (missing arms:1.331), (extra legs:1.331), (fused fingers:1.61051), (too many fingers:1.61051), (unclear eyes:1.331), bad hands, missing fingers, extra digit, bad body, pubic

上述提示词结尾引用了 3 个 Lora 模型, 需提前下载至 models/Lora 目录.

$ cd models/Lora
$ wget https://huggingface.co/datasets/KrakExilios/koreandoll/resolve/main/koreanDollLikeness_v15.safetensors
$ wget https://huggingface.co/samle/sd-webui-models/resolve/main/povImminentPenetration_ipv1.safetensors
$ wget https://huggingface.co/jomcs/NeverEnding_Dream-Feb19-2023/resolve/main/lora/breastinclassBetter_v14.safetensors

img