Free · 8 min read
Stable Diffusion for designers: how to start and what the license allows
How a designer can start with Stable Diffusion: local or cloud, which model to pick, what the license allows, how to work in ComfyUI and where it really helps.
- How Stable Diffusion differs from subscription services
- Which model to pick and what its license allows
- Your first workflow in ComfyUI: from prompt to edits
- Free
- 1,504 members
- Join in 1 click
How to start with Stable Diffusion: 5 steps
- 1
Step 1
Understand what Stable Diffusion is and how it's used
- 2
Step 2
Pick a model and read its license
- 3
Step 3
Install an interface: ComfyUI or AUTOMATIC1111
- 4
Step 4
Your first workflow: prompt, settings, edits
- 5
Step 5
Use control: that's Stable Diffusion's strength
- 1,504
- community members
- 8,915
- messages with questions since July 2024
- 9
- design games since September 2024
- 759
- people wrote in the chat
This guide is for you if you
Heard Stable Diffusion is free
And want to know what exactly is free, what needs a powerful computer and where the catches are.
Need more than just a prompt
You need control: composition from a sketch, your own style, edits to specific parts of an image.
Work with client materials
You don't want to upload layouts to someone else's service and are looking for a way to generate on your own computer.
In short
- Stable Diffusion isn't one website but a family of models from Stability AI that you can download and run yourself or in cloud services.
- The license depends on the model: SD 3.5 uses the Stability AI Community License, SDXL 1.0 uses CreativeML Open RAIL++-M. Read the model card before a commercial project.
- The Community License is free for individuals and companies with annual revenue under $1,000,000; commercial use requires registering with Stability AI.
- The easiest local start is the ComfyUI desktop app for Windows and macOS; without a powerful computer, use cloud services.
- The strength for designers is control: a sketch as the base of the composition, edits to fragments, your own style and work without uploading files.
How to start with Stable Diffusion: 5 steps
- 1
Step 1
Understand what Stable Diffusion is and how it's used
Unlike services where you pay a subscription and type a prompt on a website, Stable Diffusion is a set of models: files with neural-network "weights" that Stability AI publishes for download. To generate, a model needs a host program and a computer to run on. So there are two paths:
- locally: you install an interface (such as ComfyUI), download a model and generate on your own graphics card, with no per-image fee but with hardware requirements and setup time;
- in the cloud: you run the same models on someone else's powerful servers; for example, Comfy Cloud works in a browser on any device and charges only for running workflows, while the Stability AI API charges credits per operation (1 credit = $0.01, with 25 free credits to start);
- either way, the rules of use are set by the model's license, not by the interface you work in.
💡 If you've never done AI generation, try the cloud option first: you'll find out whether you need this level of control before spending an evening on installation.
- 2
Step 2
Pick a model and read its license
Stable Diffusion has several generations of models, and their terms differ. Stability AI's newer models, Stable Diffusion 3.5 (Large, Large Turbo, Medium) and Stable Diffusion 3 Medium, are on the Core Models list and are distributed under the Stability AI Community License. The older, still popular SDXL 1.0 base has a different license: CreativeML Open RAIL++-M. What to know about the Community License:
- research, non-commercial and commercial use is free for individuals and organizations with annual revenue under $1,000,000 (total revenue, not just from AI);
- commercial use requires registering with Stability AI; once revenue exceeds $1,000,000, the free license ends and you need an enterprise license;
- as between you and Stability AI, the images you generate are yours, to the extent your country's law allows;
- you may not use the models or their outputs to build your own "foundational" AI model; fine-tunes such as LoRAs don't count.
💡 SD 3.5 Large on Hugging Face is gated: to download the files you sign in, give your name, email and country, and accept the license. That's normal; it's what an official download should look like.
Swap workflows, not just pictures
A working ComfyUI workflow saves as a single JSON file. Share yours with the result in the community and get tips on what to simplify or add. It's free.
Share my workflowFree. 1,504 designers are already in the community.
- 3
Step 3
Install an interface: ComfyUI or AUTOMATIC1111
The most actively developed interface today is ComfyUI: a workflow is made of nodes connected by wires, a bit like a diagram. It looks unusual, but designers used to layers and components pick up the logic quickly. How to install it:
- the desktop app for Windows and macOS, which the developers call the easiest option for new users;
- a portable build for Windows, with separate versions for NVIDIA, AMD and Intel graphics cards and for CPU only;
- a manual install on Windows, Linux and macOS, supporting NVIDIA, AMD, Intel and Apple Silicon;
- the ComfyUI developers say even large models can run on 4 GB of VRAM and 8 GB of RAM, but real speed depends on your hardware, so test it on your own machine.
💡 You'll also come across AUTOMATIC1111 WebUI, the classic interface with fields and tabs that many older tutorials cover. It's still available, but its latest release (v1.10.1) came out in February 2025, so run newer models in ComfyUI.
- 4
Step 4
Your first workflow: prompt, settings, edits
Start from a ready-made template: ComfyUI has a library of template workflows for text-to-image. Put the model file in the `models/checkpoints` folder, select it in the loader node and describe what you want to see. Then go through the basic techniques one by one:
- text to image: describe the subject, technique, lighting and composition; most interfaces also have a negative prompt, a field for what shouldn't be in the frame;
- image to image (img2img): feed in your sketch or photo as the base, and the model redraws it according to the prompt;
- inpainting: mask part of the image and regenerate only that part, to remove an object, change the background or fix a hand;
- outpainting: extend the image beyond its frame, for example to turn a square into a horizontal banner;
- upscaling: a separate step once you've chosen a variant.
💡 Save a workflow that works as a JSON file: next time you start from settings you trust, not from scratch. ComfyUI can also recover the workflow and seed from supported generated files.
Stable Diffusion won't run or gives you the wrong thing? Ask the community → - 5
Step 5
Use control: that's Stable Diffusion's strength
For a quick "pretty picture", a subscription service is simpler. Stable Diffusion wins where a designer needs control:
- ControlNet: generate from a sketch, pose or depth map, so you set the composition and the model only renders it in the style you need;
- LoRA: small fine-tuned models that add a specific style or object; train them only on work you have rights to;
- local work: ComfyUI can run fully offline, since its core downloads nothing unless you ask it to, so client materials stay on your computer (still read your NDA);
- honesty in your portfolio: if part of a visual was generated, say so in the case study and show what you refined by hand.
💡 In May 2025 the community ran an AI challenge: the logo, copy and visuals for a landing page had to be generated with AI, and participants shared their prompts. A task with a deadline quickly shows where Stable Diffusion saves time and where it's easier to open Photoshop.
You're not alone with these questions
We counted what people ask about most in our community. AI tools are the fourth most popular topic, after learning, portfolios and Figma.
Messages by topic in Spaceberry Community, July 2024 — November 2026. Keyword search.
Practice with a deadline and live feedback
The best teacher is a task with a deadline that other people see. That's why the community regularly runs design games: the same brief for everyone, a deadline and a vote.
Design game · January 2026
NAOMA website redesign
The academy's home page through the eyes of an applicant, a student and a teacher. Three finalists; the community chose the winner.
AI challenge · May 2025
A landing page made by AI
Logo, copy and visuals made only with AI tools. Four entries, an anonymous vote, and participants shared their prompts.
Workshop · April 2026
Figmosha: AI in Figma
The tool's author showed how to generate interfaces in Figma and analyze a design system.
Free. 1,504 designers are already in the community.
How to join: 1 minute
- 1
Tap the button
Telegram opens with the Spaceberry community.
- 2
Join
It's free: all you need is a Telegram account.
- 3
Introduce yourself
Say where you're starting from and ask your first question.
Questions and answers
Didn't find your answer? Ask it in the community.
Is Stable Diffusion free?
The models themselves can be downloaded for free, and under the Stability AI Community License even commercial use is free for individuals and companies with annual revenue under $1,000,000 (with registration). But local generation needs your own computer with a graphics card, and cloud services and the Stability AI API are paid, by credits or running time.
Can I use generated images in commercial projects?
It depends on the specific model's license. Under the Community License (SD 3.5, SD 3 Medium), outputs are yours as between you and Stability AI, within your country's law, and commercial use requires registration. Other models, such as SDXL 1.0, have their own license: read it on the model card. Copyright law on AI images differs between countries, so for important projects consult a lawyer.
Stable Diffusion vs Midjourney: which should a designer use?
Midjourney is a subscription service: a quick start and consistently good-looking results, but less control. Stable Diffusion takes setup, but gives you sketches as a base, edits to fragments, your own styles and offline work. Many designers use both. More on Midjourney in [Midjourney for designers](/en/guides/midjourney-for-designers).
Do I need a powerful computer?
For local work, ideally a computer with a graphics card: ComfyUI supports NVIDIA, AMD, Intel and Apple Silicon. The ComfyUI developers say large models can run even on 4 GB of VRAM and 8 GB of RAM, but generation on weak hardware is slower. If your computer can't cope, run the same models in the cloud.
Is the Spaceberry community paid?
No. Spaceberry Community is free: it's supported by Spaceberry Foundation, a non-profit public organization. All you need to join is a Telegram account. The community has 1,504 designers, and AI tools are its fourth most popular topic. People share prompts and workflows, post generations for feedback and take part in design games such as the May 2025 AI challenge to build a landing page. Most conversations are in Ukrainian.
Try Stable Diffusion on a real task
Join 1,504 designers who learn, share their work and play design games. Show your generation and find out what to improve. The community talks in Ukrainian.
- Free
- 1,504 members
- Join in 1 click
Updated · Spaceberry Foundation
Sources: