安装stable diffusion过程记录
主要参考安装过程:https://nenly.notion.site/c5805e7ae26b4683a277c5586ea05904
模型下载网站:https://civitai.com/
安装顺序:1、2、7、4、5
主要遇到的难题:
1、一些博主推荐的SD基础模型用的百度盘分享,下载贼拉慢,简直受不了,所以手动去C站下载一个基础模型。
2、下载好后,上传到阿里云DSW服务器上,上传贼拉慢,被限流。所以得使用阿里云盘,挂载上去,虽然也慢,但是比直接上传服务器快些了。
3、准备执行脚本安装报错:This script must not be launched as root, aborting…
解决方法:./webui.sh -f #强制执行
4、安装SD时下载依赖贼拉慢
pip主要切换清华镜像源和阿里云镜像源,来回切换然后安装webui,需要不停尝试。按照经验,刚开始使用清华源,安装tb-nightly失败时,再切换阿里云源,继续断点续装,基本ok。
解决方法:https://www.bilibili.com/read/cv29812530/?jump_opus=1
亲测有效。
5、No matching distribution found for tb-nightly
解决方法:https://www.fujieace.com/python/error-no-matching-distribution-found-for-tb-nightly.html
https://blog.csdn.net/weixin_46455141/article/details/131353266
6、经常下载依赖中断,不断断点续装
7、creating model quickly: OSError。OSError: Can't load tokenizer for 'openai/clip-vit-large-patch14'. If you were trying to load it from 'https://huggingface.co/models', make sure you don't have a local directory with the same name. Otherwise, make sure 'openai/clip-vit-large-patch14' is the correct path to a directory containing all relevant files for a CLIPTokenizer tokenizer
解决方法:https://zhuanlan.zhihu.com/p/681188665
9、NotImplementedError: No operator found for memory_efficient_attention_forward with inputs:
原因: xformers版本问题,0.0.18版本适用于pytorch2.0,webui现在默认的版本是1.13.1,所以不兼容。可以用python -m xformers.info来查看xformers的适配情况。解决: 降低xformers版本,pip install xformers==0.0.16
原文链接:https://blog.csdn.net/Huang_Fj/article/details/130105611
最后安装的xformers==0.0.22版本
10、安装tagger
https://zhuanlan.zhihu.com/p/684871772
11、安装controlnet
https://blog.csdn.net/qq_43377653/article/details/130646734
安装插件:https://blog.csdn.net/a1112315165/article/details/134637465
操作方法和案例:https://waytoagi.feishu.cn/wiki/ZrpPwKPwji25KmkKVxmcT686nfh
提示词深入:https://baijiahao.baidu.com/s?id=1775334211764240084&wfr=spider&for=pc
8、安装yum:E: Unable to locate package yum 问题解决
真正解决方案:https://blog.51cto.com/u_16013293/6151280
//备份本地源 cp /etc/apt/sources.list /etc/apt/sources.list.bak vim /etc/apt/sources.list
#国内镜像 deb http://mirrors.aliyun.com/ubuntu/ bionic main restricted universe multiverse deb http://mirrors.aliyun.com/ubuntu/ bionic-security main restricted universe multiverse deb http://mirrors.aliyun.com/ubuntu/ bionic-updates main restricted universe multiverse deb http://mirrors.aliyun.com/ubuntu/ bionic-proposed main restricted universe multiverse deb http://mirrors.aliyun.com/ubuntu/ bionic-backports main restricted universe multiverse deb-src http://mirrors.aliyun.com/ubuntu/ bionic main restricted universe multiverse deb-src http://mirrors.aliyun.com/ubuntu/ bionic-security main restricted universe multiverse deb-src http://mirrors.aliyun.com/ubuntu/ bionic-updates main restricted universe multiverse deb-src http://mirrors.aliyun.com/ubuntu/ bionic-proposed main restricted universe multiverse deb-src http://mirrors.aliyun.com/ubuntu/ bionic-backports main restricted universe multiverse
替换完成后
sudo apt-get clean all sudo apt-get update
更新的时候报错:Err:1 http://mirrors.aliyun.com/ubuntu bionic InRelease The following signatures couldn't be verified because the public key is not available: NO_PUBKEY 3B4FE6ACC0B21F32
更换源之后需要添加公钥,执行以下命令:
注意:这里的recv-keys就是报错中的key,粘贴过来即可
公钥:3B4FE6ACC0B21F32(报错的公钥)
apt-key adv --keyserver keyserver.ubuntu.com --recv-keys 公钥
此时再次执行:
sudo apt-get update sudo apt-get install yum
开始安装yum,到了这一步就安装成功了,验证下是否可用
9、yum报错:There are no enabled repos问题原因及解决
可能是相关yum源被删除了:
wget -O /etc/yum.repos.d/CentOS-Base.repo http://mirrors.aliyun.com/repo/Centos-7.repo
yum clean all && yum makecache
http://mirrors.aliyuncs.com/centos/$releasever/os/x86_64/repodata/repomd.xm [Errno 14] HTTP Error 404 - Not Found
解决方法:http://www.manongjc.com/detail/41-upvdzrhmxrxvlyv.html
10、An error happened while trying to locate the file on the Hub and we cannot find the requested files in the local cache. Please check your connection and try again or make sure your Internet connection is on
https://zhuanlan.zhihu.com/p/684871772
总结
放弃手动安装了,各种插件的安装太痛苦了。
换了个思路,使用秋叶整合包是一样的,所有依赖都打好包了,那么直接传到linux环境不就行了吗?就这样干。
遇到的问题:
1、conda创建环境报错
配置镜像源:https://blog.csdn.net/qq_45904458/article/details/132183382
2、conda: error: argument COMMAND: invalid choice: 'activate'
https://blog.csdn.net/zhanlong11/article/details/135309140
每次使用webui都要切换空间:conda activate webui
然后安裝torch:
conda install pytorch==2.1.1 torchvision==0.16.1 torchaudio==2.1.1 pytorch-cuda=11.8 -c pytorch -c nvidia
3、下载秋叶整合包,删除git、python目录,然后分包压缩,不然包实在太大。models单独提出来压缩,总目录再压缩。
4、p7zip安装:
输入命令`sudo apt-get update`以更新系统软件包列表。
输入命令`sudo apt-get install p7zip-full`来安装7z命令。
5、上传到环境后,组装压缩包
6、到stable diffusion webui目录执行命令:python launch.py
会开始启动,过程中会自动下载依赖
ControlNet init warning: Unable to install insightface automatically. Please try run `pip install insightface` manually
注意启动时的打印报错,往往要解决,手动下载依赖等
例如下面的:
pip install datasets==2.13.0
pip install huggingface insightface huggingface_hub==0.21.4
pip install face_recognition
7、No matching distribution found for tb-nightly
解决方法:https://www.fujieace.com/python/error-no-matching-distribution-found-for-tb-nightly.html
https://blog.csdn.net/weixin_46455141/article/details/131353266
8、creating model quickly: OSError。OSError: Can't load tokenizer for 'openai/clip-vit-large-patch14'. If you were trying to load it from 'https://huggingface.co/models', make sure you don't have a local directory with the same name. Otherwise, make sure 'openai/clip-vit-large-patch14' is the correct path to a directory containing all relevant files for a CLIPTokenizer tokenizer
解决方法:https://zhuanlan.zhihu.com/p/681188665
9、Could not locate model-keyword extension
重新安装model-keyword扩展,安装好就不会出现这个提示
10、安裝tagger会遇到的问题
huggingface_hub.utils._errors.LocalEntryNotFoundError: An error happened while trying to locate the file on the Hub and we cannot find the requested files in the local cache. Please check your connection and try again or make sure your Internet connection is on.
注意,修改完成后,重新执行python launch.py
https://zhuanlan.zhihu.com/p/684871772
11、使用easyphoto训练
easyphoto训练:https://cloud.tencent.com/developer/article/2358570
https://zhuanlan.zhihu.com/p/681308962
遇到的问题,主要是勾选validation会碰见unet验证报错:INFO - __main__ - Running validation error, skip it.Error info: UNet2DConditionModel.forward() got an unexpected keyword argument 'added_cond_kwargs'.
解决方法:加一行代码
进阶:
小红书instanceID部署:https://www.bilibili.com/read/cv31000540
官方源址:https://github.com/InstantID/InstantID
官方模型下载:https://mp.weixin.qq.com/s/qpMvZajNI2WovyymALAajQ
下载lcm-lora-sdxl模型:https://huggingface.co/latent-consistency/lcm-lora-sdxl/tree/main
if not USE_PEFT_BACKEND:
raise ValueError("PEFT backend is required for this method.")
解决方法:https://blog.csdn.net/weixin_43457608/article/details/136060188
运行instanceID爆显存: If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation
把图片size改小,不然爆显存。
安装ComfyUI:https://blog.csdn.net/weixin_42735060/article/details/136096152
mv extra_model_paths.yaml.example extra_model_paths.yaml
工作流下载:https://comfyworkflows.com/workflows/ce41e3cc-7a01-4f01-ae5e-8c69145e4521
ComfyUI使用instanceID:https://zhuanlan.zhihu.com/p/682107843
记录自己工作的技术积累