Mastering Your Renderings: A Step-by-Step Guide to Installing Stable Diffusion on Windows
Mastering Your Renderings: A Step-by-Step Guide to Installing Stable Diffusion on Windows
AI
10 minutes
Aug 10, 2024
How to install Stable Diffusion on your PC
How to install Stable Diffusion on your PC
So you want some piece of AI on your PC? Welcome, buddy!
In this post, you'll find a helpful guide on "How to install Stable Diffusion on your PC"
First things first
Architects who have not yet committed to bespoke photorealistic 3D renders may find Stable Diffusion useful.
Using AI can help you to make your website's renderings more consistent in terms of the quality of visuals to prevent inconsistencies. What do I mean by inconsistencies? Well, you made a great decision by hiring a professional photographer to capture your completed projects, right? But then, you did not make the same decision about your ongoing projects' renders. Sorry for the harsh truth :/ You can now at least experiment with AI and use it for brainstorming or increasing the quality of your 3D visuals. Let's jump in.
Meet the requirements (No, it's not a dating app)
First things first, let’s make sure your PC isn’t a relic from the era when dial-up was a thing. You’ll need a relatively modern machine with a decent GPU. If you're not sure what a GPU is, it's the thing on your computer that stops Minecraft from looking like a slideshow. It's also a good idea to run Windows 10 or 11 rather than Windows 95, as Windows 95 won't do.
Here are those bullet points you're dying to see;
Windows 10 or higher
Discrete Nvidia video card (GPU) with at least 4 GB VRAM
16GB of RAM
At least 10GB of disk space
So you want some piece of AI on your PC? Welcome, buddy!
In this post, you'll find a helpful guide on "How to install Stable Diffusion on your PC"
First things first
Architects who have not yet committed to bespoke photorealistic 3D renders may find Stable Diffusion useful.
Using AI can help you to make your website's renderings more consistent in terms of the quality of visuals to prevent inconsistencies. What do I mean by inconsistencies? Well, you made a great decision by hiring a professional photographer to capture your completed projects, right? But then, you did not make the same decision about your ongoing projects' renders. Sorry for the harsh truth :/ You can now at least experiment with AI and use it for brainstorming or increasing the quality of your 3D visuals. Let's jump in.
Meet the requirements (No, it's not a dating app)
First things first, let’s make sure your PC isn’t a relic from the era when dial-up was a thing. You’ll need a relatively modern machine with a decent GPU. If you're not sure what a GPU is, it's the thing on your computer that stops Minecraft from looking like a slideshow. It's also a good idea to run Windows 10 or 11 rather than Windows 95, as Windows 95 won't do.
Here are those bullet points you're dying to see;
Windows 10 or higher
Discrete Nvidia video card (GPU) with at least 4 GB VRAM
16GB of RAM
At least 10GB of disk space
Install - Install - Install
Install - Install - Install
Step 1: Install Python (No, we do not code anything, chill)
You have to install Python 3.10.6. (DON’T use Python 3.11 or newer)
Visit the Python website and download the Windows installer (64-bit).
Once you start installing, make sure to select "Add Python to 3.10 to PATH".
Step 2: Install git (You know the drill)
Download and install git from this website.
Step 3: Download the Stable Diffusion (Finally, right?)
Click the Windows key, search for "cmd" and run it. You should see a black window open. If you're not feeling like a hacker already, you may just open the wrong thing.
Now copy what you see below and paste it into the command window.
cd %userprofile%
You should see your prompt shows something like C:\Users\YOUR_NAME>
Now copy what you see below and paste it into the command window. This will download/install Stable Diffusion WebUi
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
A folder called stable-diffusion-webui will be created in your home directory.
Now copy what you see below paste it into the address bar and press enter.
%userprofile%\stable-diffusion-webui
Now locate the "webui-user.bat file inside this folder, right-click, and edit it with the notepad.
Now copy what you see below and paste it where it says "set COMMANDLINE_ARGS="
set COMMANDLINE_ARGS=--xformers --medvram
You should have correctly installed and configured the Stable Diffusion.
Optional
If you want a shortcut on your desktop for easier access, copy the "webui-user.bat" file and paste it on your desktop.
If something is wrong with you (like me), click here and download an icon for proper appearance.
Right-click on the "webui-user.bat" file, > properties > change icon > select the icon file you've just downloaded.
Step 4: Download the necessary models and everything else (bear with me)
Once you run the "webui-user.bat" file (you can change its name to "Stable Diffusion", you'll see something very similar to this window below. (It'll automatically open a website with think address "http://127.0.0.1:7860/"
Step 1: Install Python (No, we do not code anything, chill)
You have to install Python 3.10.6. (DON’T use Python 3.11 or newer)
Visit the Python website and download the Windows installer (64-bit).
Once you start installing, make sure to select "Add Python to 3.10 to PATH".
Step 2: Install git (You know the drill)
Download and install git from this website.
Step 3: Download the Stable Diffusion (Finally, right?)
Click the Windows key, search for "cmd" and run it. You should see a black window open. If you're not feeling like a hacker already, you may just open the wrong thing.
Now copy what you see below and paste it into the command window.
cd %userprofile%
You should see your prompt shows something like C:\Users\YOUR_NAME>
Now copy what you see below and paste it into the command window. This will download/install Stable Diffusion WebUi
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
A folder called stable-diffusion-webui will be created in your home directory.
Now copy what you see below paste it into the address bar and press enter.
%userprofile%\stable-diffusion-webui
Now locate the "webui-user.bat file inside this folder, right-click, and edit it with the notepad.
Now copy what you see below and paste it where it says "set COMMANDLINE_ARGS="
set COMMANDLINE_ARGS=--xformers --medvram
You should have correctly installed and configured the Stable Diffusion.
Optional
If you want a shortcut on your desktop for easier access, copy the "webui-user.bat" file and paste it on your desktop.
If something is wrong with you (like me), click here and download an icon for proper appearance.
Right-click on the "webui-user.bat" file, > properties > change icon > select the icon file you've just downloaded.
Step 4: Download the necessary models and everything else (bear with me)
Once you run the "webui-user.bat" file (you can change its name to "Stable Diffusion", you'll see something very similar to this window below. (It'll automatically open a website with think address "http://127.0.0.1:7860/"
More Installation
More Installation
More Installation
More Installation
Installing ControlNet
Option 1: When you are on the web UI interface, go to > Extensions > Available and then click on "Load from" and search for "sd-webui-controlnet" > click "install"
Option 2: When you are on the web UI interface, go to > Extensions > Install from URL and then paste the link below and install.
https://github.com/Mikubill/sd-webui-controlnet
No matter what option you choose, go to > Extensions > Installed > Apply and restart UI
Download & Install ControlNet Model(s)
Go to this page: https://huggingface.co/lllyasviel/ControlNet-v1-1/tree/main
Download the "control_v11f1p_sd15_depth.pth" file and move it to the folder:
stable-diffusion-webui\extensions\sd-webui-controlnet\models
Download & Install Stable Diffusion Checkpoint Model(s)
Visit: https://civitai.com/
Download the following checkpoint models;
CyberRealistic:
https://civitai.com/models/15003/cyberrealistic
Realistic Vision:
https://civitai.com/models/4201/realistic-vision-v60-b1
epiCPhotoGasm:
https://civitai.com/models/132632/epicphotogasm
Now cut and move all these files to the folder: stable-diffusion-webui\models\Stable-diffusion
Download & Install Lora Model(s)
Architecture Exterior:
https://civitai.com/models/25537/xsarchitectural-11fantasyarchitecture
You can check out more Lora models and download what you like. Here are a few examples you can check out.
Now cut and move all these files to the folder: stable-diffusion-webui\models\Lora
Installing ControlNet
Option 1: When you are on the web UI interface, go to > Extensions > Available and then click on "Load from" and search for "sd-webui-controlnet" > click "install"
Option 2: When you are on the web UI interface, go to > Extensions > Install from URL and then paste the link below and install.
https://github.com/Mikubill/sd-webui-controlnet
No matter what option you choose, go to > Extensions > Installed > Apply and restart UI
Download & Install ControlNet Model(s)
Go to this page: https://huggingface.co/lllyasviel/ControlNet-v1-1/tree/main
Download the "control_v11f1p_sd15_depth.pth" file and move it to the folder:
stable-diffusion-webui\extensions\sd-webui-controlnet\models
Download & Install Stable Diffusion Checkpoint Model(s)
Visit: https://civitai.com/
Download the following checkpoint models;
CyberRealistic:
https://civitai.com/models/15003/cyberrealistic
Realistic Vision:
https://civitai.com/models/4201/realistic-vision-v60-b1
epiCPhotoGasm:
https://civitai.com/models/132632/epicphotogasm
Now cut and move all these files to the folder: stable-diffusion-webui\models\Stable-diffusion
Download & Install Lora Model(s)
Architecture Exterior:
https://civitai.com/models/25537/xsarchitectural-11fantasyarchitecture
You can check out more Lora models and download what you like. Here are a few examples you can check out.
Now cut and move all these files to the folder: stable-diffusion-webui\models\Lora
Finally, you are ready to give it a try to Stable Diffusion (AUTOMATIC1111)
Once you're finished with moving files around, re-start web-ui.
I'd like to use the checkpoint models merged. Here is how you can do it too.
Once you're finished, select the merged model and go to the img2img tab.
Once you put some prompts here and there, now adjust the following settings:
Don't forget to click on the "ruler" icon. Do not use high-resolution images. Maybe below 1500px?
Sampling steps could be around 20-40 or more if you have a high-end GPU.
Denoising strength determines how crazy AI should go. The lower means your prompt is more important, and vice versa.
The last (I mean it) steps are the following:
Once you're finished with moving files around, re-start web-ui.
I'd like to use the checkpoint models merged. Here is how you can do it too.
Once you're finished, select the merged model and go to the img2img tab.
Once you put some prompts here and there, now adjust the following settings:
Don't forget to click on the "ruler" icon. Do not use high-resolution images. Maybe below 1500px?
Sampling steps could be around 20-40 or more if you have a high-end GPU.
Denoising strength determines how crazy AI should go. The lower means your prompt is more important, and vice versa.
The last (I mean it) steps are the following:
Now all you have to do is experiment!
Now all you have to do is experiment!
I put one of my 5-year-old renderings to the SD.
Here is my outcome.
Spot the artifacts? I'm sure you did.
You can not control everything with AI.
Now, don't get me wrong, Stable Diffusion is pretty impressive. But like any good brainstorming session, it can get a little...unpredictable. You may have something interesting for sure but not a thing that is client-approved.
Hear the drums??? A soft CTA is coming your way!
IMO, AI can be used to experiment, try different ideas, and brainstorm. You DO NOT have full control over it. It's not like I do with 3ds Max, Corona Render, and Photoshop.
Now give your designs what they deserve. Ditch the sketches, and basic Lumion or Enscape renders, and have photorealistic renderings for your premium designs.
Please go check out my work here, and get in touch today!
I put one of my 5-year-old renderings to the SD.
Here is my outcome.
Spot the artifacts? I'm sure you did.
You can not control everything with AI.
Now, don't get me wrong, Stable Diffusion is pretty impressive. But like any good brainstorming session, it can get a little...unpredictable. You may have something interesting for sure but not a thing that is client-approved.
Hear the drums??? A soft CTA is coming your way!
IMO, AI can be used to experiment, try different ideas, and brainstorm. You DO NOT have full control over it. It's not like I do with 3ds Max, Corona Render, and Photoshop.
Now give your designs what they deserve. Ditch the sketches, and basic Lumion or Enscape renders, and have photorealistic renderings for your premium designs.
Please go check out my work here, and get in touch today!