Install Forge and Generate Your First AI Image

How to Install Forge and Generate Your First AI Image

If you’ve been exploring local AI, you’ve probably heard names like Stable Diffusion, Flux, SDXL, Forge, ComfyUI, checkpoints, and LoRAs being thrown around everywhere.

For beginners, it can feel overwhelming because many people talk about the models without explaining how you actually use them.

One of the most common misconceptions is that Stable Diffusion is a piece of software you install on your computer.

That’s only partially true.

Stable Diffusion is actually a family of AI image generation models. To use those models, you need a piece of software that loads them and provides an interface where you can write prompts, adjust settings, and generate images.

That’s where Forge comes in.

Forge is one of the easiest ways to start generating AI images locally. It provides a browser-based interface that allows you to run Stable Diffusion, Flux, and many other image generation models directly on your own hardware.

In my What Is Stable Diffusion? guide, we covered the technology that powers many of today’s AI image generation models. In this tutorial, we’re taking the next step by installing Forge and generating your first image locally.

By the end of this guide, you’ll have Forge installed, a model loaded, and your first AI-generated image running on your own machine.

What Is Forge?

Stable Diffusion Forge is a local AI image generation application built on top of the Stable Diffusion WebUI ecosystem.

Think of Forge as the control panel that allows you to interact with AI image generation models.

A common beginner mistake is confusing Forge with Stable Diffusion or Flux.

They’re related, but they’re not the same thing.

Component	Purpose
Forge	The application and user interface
Stable Diffusion / Flux	The AI image generation model
Checkpoint	The model file loaded into Forge
Prompt	Your instructions for the model
LoRA	A small add-on that teaches styles, characters, or concepts

If you’ve used Ollama before, the relationship is very similar.

Ollama is the software that runs local language models.

Forge is the software that runs local image generation models.

The model itself still needs to be downloaded separately.

Why Use Forge?

There are several ways to generate AI images locally today, but Forge has become one of the most popular starting points because it strikes a nice balance between simplicity and capability.

Unlike cloud-based image generation platforms, Forge runs directly on your own computer. Once everything is installed, you can generate images without subscription fees, token costs, or relying on a third-party service. You’re in complete control of your models, settings, and outputs.

What makes Forge especially appealing for beginners is that it removes much of the complexity that comes with more advanced image generation tools. You still get access to powerful features such as Stable Diffusion and Flux models, LoRAs, inpainting, image-to-image generation, and model management, but through a simple browser-based interface that’s easy to understand.

For most people, Forge is the quickest path from downloading a model to generating a usable image. Instead of spending hours learning workflow graphs and node connections, you can focus on learning prompts, experimenting with models, and understanding how AI image generation works.

That’s one of the reasons I recommend starting with Forge before moving into more advanced tools like ComfyUI. Once you’re comfortable generating images and understand the basics, transitioning into more complex workflows becomes much easier.

Forge vs Stable Diffusion

This is where many beginners get confused.

Forge and Stable Diffusion are not competitors. They solve different problems.

Stable Diffusion	Forge
The AI image generation model	The software that runs the model
Creates images	Provides the interface
Downloaded as a checkpoint file	Installed as an application
Different models available	One interface that can run many models

You can think of Stable Diffusion as the engine and Forge as the dashboard that lets you drive it.

Forge can also run newer models such as Flux, meaning that once you learn Forge, you’re not limited to a single model family.

What You’ll Accomplish In This Guide

By the end of this tutorial, you’ll have:

Installed Forge
Downloaded your first image generation model
Loaded the model into Forge
Generated your first AI image locally
Learned the basic settings needed to start experimenting

Now that you understand what Forge is and how it fits into the local AI ecosystem, let’s get it installed.

System Requirements

Before installing Forge, it’s important to understand that AI image generation is heavily dependent on your hardware.

The more powerful your GPU and the more VRAM available, the larger and more capable models you’ll be able to run.

If you’re just getting started, don’t worry about running the biggest model available. Focus on getting Forge installed and generating your first image successfully.

Hardware	Recommended Models
4GB–8GB VRAM	Stable Diffusion 1.5
8GB–12GB VRAM	SDXL
12GB–24GB VRAM	SDXL, Flux Schnell
16GB–24GB+ VRAM	Flux Dev and larger models

Mac users can also run Forge, although setup can require additional troubleshooting compared to Windows installations.

How To Install Forge On Windows

For most users, Windows is the easiest platform for installing Forge.

Start by visiting the official Forge GitHub repository.

Official Forge Repository:
https://github.com/lllyasviel/stable-diffusion-webui-forge

Scroll down until you find the installation section and download the latest one-click installer package.

After the download finishes:

Extract the downloaded archive.
Open the extracted Forge folder.
Run update.bat.
Wait for Forge to download and update any required files.
After the update finishes, run run.bat.

The first launch may take several minutes while Forge installs Python packages and dependencies.

When installation completes, Forge will automatically launch in your browser.

In most cases the local address will be:

http://127.0.0.1:7860

If your browser doesn’t open automatically, copy that address into your browser manually.

Installing Forge On macOS

Forge can run on macOS, but installation is currently more involved than the Windows version.

You’ll first need Homebrew, Git, and Python installed.

brew install git python@3.10

Next, create a folder for Forge and clone the repository.

mkdir -p ~/AI
cd ~/AI
git clone https://github.com/lllyasviel/stable-diffusion-webui-forge.git
cd stable-diffusion-webui-forge

Edit the webui-user.sh file and point it to Python 3.10.

python_cmd="/opt/homebrew/bin/python3.10"

Many Apple Silicon users also add the following launch arguments during initial setup.

export COMMANDLINE_ARGS="--skip-torch-cuda-test --no-half --use-cpu interrogate"

Then launch Forge using:

./webui.sh

Once startup completes, open:

http://127.0.0.1:7860

As of writing, macOS support can require some troubleshooting depending on your hardware and Python environment. If you run into issues, don’t panic. The local AI ecosystem changes quickly, and newer installation methods continue to improve.

Installing Forge On Linux

Linux users can install Forge directly from the GitHub repository.

sudo apt update
sudo apt install git python3 python3-venv python3-pip

mkdir -p ~/AI
cd ~/AI
git clone https://github.com/lllyasviel/stable-diffusion-webui-forge.git
cd stable-diffusion-webui-forge

Launch Forge using:

./webui.sh

Most NVIDIA users will want current NVIDIA drivers installed before attempting image generation.

Once Forge is running, the next step is downloading and installing your first image generation model.

Downloading Your First Model

At this point, Forge is installed and running, but you’re not quite ready to generate images yet.

Forge is the application. It still needs a model before it can create images.

If you click the checkpoint dropdown in a fresh Forge installation, you’ll usually see that no models are available.

This is normal.

The next step is downloading your first model checkpoint.

Where To Download Models

The two most popular places to download image generation models are Hugging Face and Civitai.

Hugging Face is generally the best source for official releases and research models.
Civitai is the largest community repository for checkpoints, LoRAs, styles, and image generation resources.

For beginners, I recommend starting with an official model before exploring community-created variations.

Model	Best For
SDXL Base 1.0	Official beginner model
Stable Diffusion 1.5	Older hardware and lower VRAM systems
Juggernaut XL	Realistic images
RealVisXL	Photorealistic content
DreamShaper XL	General-purpose art and photography

If this is your first time using Forge, start with SDXL Base 1.0 or another beginner-friendly SDXL checkpoint.

Where To Put Models In Forge

Most checkpoints are downloaded as .safetensors files.

Once downloaded, place the file into Forge’s Stable Diffusion model folder.

Windows

Forge\models\Stable-diffusion\

macOS

~/AI/stable-diffusion-webui-forge/models/Stable-diffusion/

Linux

~/AI/stable-diffusion-webui-forge/models/Stable-diffusion/

After copying the model into the folder, restart Forge or click the refresh button next to the checkpoint selector.

Your model should now appear in the checkpoint dropdown.

Generate Your First AI Image

Now comes the fun part.

Open the txt2img tab.

This mode allows you to generate an image using only a text prompt.

For your first test image, try the following prompt:

a cinematic photo of a cozy cabin in a snowy forest at night, warm light glowing from the windows, soft falling snow, detailed, realistic, beautiful composition

If your model supports negative prompts, enter:

blurry, low quality, distorted, bad anatomy, extra limbs, text, watermark, logo, oversaturated, ugly, deformed

Use these beginner-friendly settings:

Sampler: DPM++ 2M Karras
Steps: 25–30
CFG Scale: 6–7
Width: 1024
Height: 1024
Batch Count: 1
Batch Size: 1
Seed: -1

Click Generate.

After a short wait, Forge will produce your first AI-generated image.

Congratulations. You’ve officially generated your first image locally using Forge.

If the image isn’t exactly what you wanted, that’s completely normal. AI image generation is an iterative process. Small prompt changes can produce dramatically different results.

Understanding The Most Important Settings

You don’t need to understand every option immediately, but these are the settings you’ll use most often.

Setting	Purpose
Prompt	What you want to create
Negative Prompt	What you want to avoid
Steps	How many refinement passes the model performs
CFG Scale	How strongly the model follows your prompt
Sampler	The image generation method
Seed	The starting randomness used to create the image
Width / Height	The image resolution

For now, don’t worry about mastering every setting. Focus on generating images and experimenting with prompts.

Once you’re comfortable with the basics, you can begin exploring LoRAs, image-to-image workflows, inpainting, upscalers, and eventually more advanced tools like ComfyUI.

Final Thoughts

Getting started with local AI image generation is much easier than most people think.

Forge provides one of the simplest ways to install a model, enter a prompt, and begin generating images on your own hardware.

My biggest piece of advice is to keep things simple at first.

Install Forge
Download one model
Generate one image
Learn prompts
Experiment

Don’t worry about installing ten LoRAs, experimenting with advanced workflows, or learning every setting on day one.

Once you can consistently generate images, you’ll have a solid foundation for everything else in the local AI image generation ecosystem.

Common Forge Installation Problems

If Forge doesn’t launch successfully on your first attempt, don’t panic. Most installation issues are caused by missing dependencies, incorrect model placement, or hardware limitations.

Forge Opens But No Models Appear

This is usually caused by placing your checkpoint in the wrong folder.

Verify that your model file is located inside the Stable Diffusion models directory and click the refresh button beside the checkpoint selector.

Out Of Memory Errors

If you receive out-of-memory errors, your GPU may not have enough VRAM for the selected model or resolution.

Reduce image resolution
Lower batch size
Use a smaller model
Try Stable Diffusion 1.5 instead of SDXL or Flux

Forge Will Not Launch On macOS

macOS installations can be more sensitive to Python versions and dependency conflicts than Windows installations.

Make sure Python 3.10 is installed and referenced correctly in your Forge configuration files.

If you have any other difficulties consider a python virtual environment to make sure you have no conflicts.

Images Generate But Look Bad

This is usually not a Forge problem.

Most often the issue is:

A weak prompt
Poor model selection
Incorrect settings
Expecting too much from the first generation

AI image generation is an iterative process. Small prompt adjustments can dramatically improve results.

Frequently Asked Questions

Is Forge Free?

Yes. Forge is free and open source. Once installed, you can generate images locally without paying subscription fees or token costs.

Can Forge Run Flux Models?

Yes. Forge supports Flux along with Stable Diffusion 1.5, SDXL, and many other image generation models.

Do I Need An Internet Connection?

Only for downloading Forge and model files. Once everything is installed, image generation happens locally on your machine.

Should I Use Forge Or ComfyUI?

For beginners, Forge is usually the easiest place to start.

Once you’re comfortable generating images and understand how models work, you can move into ComfyUI for more advanced workflows and automation.

What Model Should Beginners Start With?

SDXL Base 1.0 is a great starting point because it’s an official model with excellent documentation and a large community.

Where To Go Next

Now that Forge is installed and you’ve generated your first image, you’re ready to explore the wider local AI image generation ecosystem.

I hope this guide helped you get Forge installed and generate your first AI image.

Local AI image generation can seem intimidating at first, but once you get your first model running and create your first image, the entire ecosystem starts making much more sense.

Start simple, experiment often, and don’t worry about mastering every setting on day one.

Stay sharp,
Michael Stables
Creator of GetPrompting

Enjoying the content?

GetPrompting is independently run, and I’m keeping the tutorials, guides, and workflow experiments free.

If you’d like to support future content, you can buy me a coffee.

Buy Me a Coffee

Totally optional. The site stays free either way.