By the end of this guide you will have turned a written description into a finished image and saved it in a format you can use for a post, a presentation or a design. The job is done by an AI text-to-image generator: you describe what you want in a prompt, the model draws it, and you refine the result with edits or a revised prompt.
If you want to generate images from text with AI, the real decisions are how much you want to pay, whether the image needs readable text, and whether you need commercial rights or privacy. This guide compares seven tools from our Image generation category.
Key takeaways
- Write a clear prompt (subject, setting, style), generate several variations, pick one, refine it with edits or a tweaked prompt, then export it.
- The main decision is how you access the model: a paid subscription, a free web tool with limits, an app inside a design suite, or a local app on your own computer.
- Free options are limited: for example 30 credits a week on Ideogram and 150 fast tokens a day on Leonardo.Ai, and Midjourney has no free plan.
- Check licensing and privacy: some free plans publish your creations by default, and commercial use can depend on the plan.
What you need before you start
- A clear idea. One sentence about the subject, the setting and the look you want is enough to begin.
- An account. Most tools need a sign-up. Some web tools work without one, and a local app needs only a download.
- A budget. Decide whether you will use a free tier, a monthly plan or pay-as-you-go credits.
- Time. Your first image can take minutes. Getting one you are happy with usually takes several rounds and under an hour.
- Difficulty. Easy for the web tools.
- Rights and care. Check the licence for commercial use. Avoid prompts that imitate a living artist's work or a real person's likeness without permission, and do not present generated images as photographs of real events.
Which kind of AI tool fits this job
Pick by how you want to work and what you plan to do with the image.
A subscription generator with a member gallery. Midjourney creates images and short videos from written prompts and is subscription only, with Basic, Standard, Pro and Mega tiers. You can use it on the website, in Discord, or on iOS and Android.
An app inside a design toolset. Adobe Firefly generates images and also lets you fill, remove and expand parts of a picture, and choose Adobe's own models or partner models. Adobe says its content is designed to be commercially safe and is not used to train models.
A generator tuned for readable text. Ideogram focuses on designs that include legible typography, and adds prompt-based editing, image reference and character consistency. Reve Image plans the layout before it renders, lets you edit by asking, and lists placement of text inside compositions.
A platform with many models and training. Leonardo.Ai offers more than 80 models on paid plans, a free daily token plan and personal model training for subscribers.
Free or offline options. Stable Diffusion AI runs Stable Diffusion in the browser for free, with rate limits. CHARL-E is a free Mac app that runs Stable Diffusion locally, so nothing is sent to its servers after the first install.
Step 1: Define what the image is for
A purpose makes the prompt sharper, so decide it first.
- Write down where the image will appear: a blog header, a slide, a poster, a social post.
- Note the shape you need, such as square, portrait or landscape. Stable Diffusion AI lists square, portrait and landscape formats.
- Decide if it needs words in the image. If so, favour a tool built for text, such as Ideogram or Reve Image.
- Decide whether it is for commercial use, since that can change which plan you need.
Check it worked by being able to finish the sentence: "I need an image of ... for ... in ... format."
Step 2: Choose a tool and plan
You finish this step with one tool ready.
- Match the tool to your constraint: free web use, subscription quality, a design suite, or offline privacy.
- Check the free tier. Ideogram's free accounts get 30 credits a week, Leonardo.Ai's free plan allows 150 fast tokens per day, and Stable Diffusion AI is free with rate limits.
- Check public-by-default rules. Leonardo.Ai says Free plan creations are public, and Ideogram says images are published to the community gallery by default unless you have a Plus or Pro plan.
- Use the comparison table further down to see pricing and stated limits side by side.
Check it worked by knowing whether your images will be private and whether the plan covers your intended use.
Step 3: Write a first prompt
The goal is a prompt specific enough to guide the model without overloading it.
- Start with the subject: what is in the picture and what it is doing.
- Add the setting, lighting and mood, for example "on a wooden table in soft morning light".
- Add the style, such as photograph, watercolour, flat illustration or 3D render.
- If the tool has controls, use them for the details you care about. Stable Diffusion AI lists negative prompt, CFG scale and seed controls.
Check it worked by reading the prompt aloud and confirming a stranger could picture the same scene. For more on wording, see how to write prompts for AI image generators.
Step 4: Generate several variations
Treat the first result as a draft and look at more than one.
- Run the prompt and review every image returned.
- Generate again with small changes rather than rewriting everything.
- Keep notes on which wording changed the result.
- Watch your allowance. Credits and tokens are metered on most tools, and unused allowances may not roll over: Ideogram says unused monthly credits do not, and Midjourney says unused Fast GPU time does not.
Check it worked by having two or three candidates that are close to what you imagined.
Step 5: Refine and edit
Fix problems with edits before you start over.
- Look for typical flaws: odd hands, garbled text, wrong proportions, objects in the wrong place.
- Use the editing tools. Adobe Firefly lists generative fill, remove and expand, Ideogram lists prompt-based editing, and Reve Image lets you edit by asking.
- Try an image reference if you want to match a look. Ideogram lists image reference and Firefly lists reference images.
- If you need the same character across images, check for a consistency feature. Ideogram lists character consistency on paid plans.
Check it worked by comparing the edited image against your purpose from Step 1.
Step 6: Upscale and export
Save the image at a size that suits where it will be used.
- Export at the highest quality your plan allows. Ideogram lists quality export on its Plus plan and above, and Reve Image says its renderer works at native 4K x 4K resolution.
- If the image is too small, use an upscaler. Stable Diffusion AI lists upscaling up to 2x.
- Save the prompt with the file so you can recreate or adjust it.
- Keep a record of the tool and plan you used, in case you need to show your licence.
Check it worked by viewing the exported file at the final size.
Step 7: Check rights and disclose where needed
Before you publish, think about permission and honesty.
- Read the commercial terms of your plan. Midjourney says a company above $1,000,000 USD yearly gross revenue must use Pro or Mega for commercial use of its output.
- Check privacy terms. Reve Image's terms allow Reve to use content for model training unless a paid account opts out.
- Do not generate or publish images that misrepresent real people or events.
- Label AI images where a platform, client or publication asks for it.
Check it worked by being able to say which licence covers the image and where it came from.
Tools for this job compared
| Tool | Good fit for | Pricing | Stated limits |
|---|---|---|---|
| Midjourney | Subscription artwork on web, Discord, iOS and Android | Paid, from $10/mo (checked 24 Oct 2026) | No free plan; unused Fast GPU time does not roll over; Stealth mode only on Pro and Mega |
| Adobe Firefly | Generate, fill, expand and vector work with Adobe or partner models | Freemium, paid from $9.99/mo (checked 23 Oct 2026) | Generative credits are metered; premium features such as video use more credits; some tools in beta |
| Ideogram | Images with readable text, editing and batch generation | Freemium, paid from $15/mo (checked 24 Oct 2026) | Free accounts get 30 credits per week; images public by default unless Plus or Pro; credits do not roll over |
| Leonardo.Ai | Many models, personal model training and teams | Freemium, paid from $12/mo (checked 24 Oct 2026) | Free plan allows 150 fast tokens per day and only the Auto model; Free plan creations are public |
| Reve Image | Layout planning, editing by request and 4K output | Freemium, paid from $7.99/mo (checked 24 Oct 2026) | Free plan has a smaller image allowance; video needs Pro; content may be used for training unless a paid account opts out |
| Stable Diffusion AI | Free browser generation with inpainting | Free | No-account UI fixed at Stable Diffusion 1.5 and 512 px; shared GPU; rate limited |
| CHARL-E | Private, offline generation on a Mac | Open source | Mac only; Intel Mac support in progress; few settings; speed depends on your Mac |
Prices and plan limits change, so confirm the current terms on each vendor's site before you subscribe.
Common mistakes to avoid
- Writing a vague prompt. "A nice picture of a dog" leaves the model to guess everything. Fix it by naming the subject, setting, lighting and style.
- Expecting readable text from every tool. Many generators garble lettering. Fix it by using a tool that lists text as a strength, such as Ideogram, and by proofreading any words in the image.
- Publishing on a free plan without checking visibility. Some free plans make creations public by default. Fix it by reading the plan terms and upgrading, or by using a local option, before generating anything private or client-related.
- Ignoring commercial terms. A plan that suits a hobby may not cover a company's use. Fix it by checking the licence before using output in ads, products or client work.
- Burning through credits on one prompt. Metered allowances run out and may not roll over. Fix it by testing ideas at low cost first and keeping notes on what works.
How we put this together
This is an evidence-based guide built from each tool's listing in our directory, which is compiled from the vendor's own site and pricing page. It is not a review based on using the tools. We don't rank tools, and nobody paid to be included. Some outbound links may be affiliate links; see our disclosure.
Frequently asked questions
Is there a free way to generate images from text with AI?
Yes, with limits. Stable Diffusion AI is free with rate limits, Ideogram free accounts get 30 credits a week, Leonardo.Ai's free plan gives 150 fast tokens per day, and CHARL-E is a free Mac app. Midjourney has no free plan. Free plans may come with rate limits, small allowances or public galleries.
Can I use AI-generated images commercially?
It depends on the tool and plan. Adobe says Firefly content is designed to be commercially safe, Midjourney requires larger companies to use Pro or Mega, and Stable Diffusion AI and Reve Image have their own terms. Read the licence for your plan before using images in ads or products.
Which tool is good for generating images on a phone?
Midjourney lists iOS and Android alongside web and Discord. Other tools here are listed for the web or, in CHARL-E's case, for Mac only. Check each vendor's site for current app options.
Can I keep my images private?
Some plans keep work private and others do not. Midjourney offers Stealth mode on Pro and Mega, Ideogram offers private generation on Plus and Pro, and Leonardo.Ai says Free plan creations are public. CHARL-E runs locally and says nothing is sent to its servers.
How long does it take to generate an image?
Stable Diffusion AI says you get an image back in seconds, and CHARL-E's author notes that local generation speed depends on your Mac and can be slow. Plan on several rounds of prompting and editing; as a rough estimate, that often fits in under an hour.
Compare more options in our Image generation category.
