What you’ll be able to do

By the end of this you’ll have a poster that actually works. Words spelled right, layout that guides the eye where it needs to go, and a file you can print or stick on Instagram without squinting at it later and spotting the date’s wrong. That’s the whole job, really. Sounds simple. It’s not always, but we’ll get there.

Before you start

Right, here’s the bit nobody tells you before you dive in. AI image generators are genuinely bad at spelling. Not a little bad, properly bad. You ask for a poster that says “SUMMER FESTIVAL, 14 JUNE” and you might get “SUMMR FESTIVL” or the date rendered as some nonsense that looks like letters but isn’t. This has got better over time but it hasn’t gone away, and I don’t think it will overnight.

So the trick, and this is the whole secret of this guide really, is to stop asking the image model to be a typesetter. It’s not one. It’s brilliant at mood, colour, composition, vibe. It’s rubbish at precise text. The workaround almost everyone uses now is to generate the artwork with AI, the background, the illustration, the feel of the thing, and then add your actual words in a proper design tool afterwards. More on that below.

The other beginner mistake is layout. People ask for “a poster for my gig” and get something busy, cluttered, with no clear order of importance. A good poster has hierarchy: one big thing (what’s happening), one medium thing (when and where), and small print for the rest (ticket links, sponsors, whatever). If your poster doesn’t have that structure, nobody reads past the first three seconds.

Get set up

For most people making a poster, I’d point you at Canva first. It’s free to start, it’s built for exactly this, and it has templates plus a “Magic Design” feature that takes a prompt or an image and suggests layouts you can then edit properly, with real text boxes you can fix typos in. That’s the practical route for almost everyone reading this.

If you want AI-generated artwork with text baked in and you’re willing to accept it needs checking, Ideogram is worth knowing about. It’s built specifically to render text inside images more reliably than most other generators, so if you want your poster’s headline actually inside the AI image rather than added afterwards, that’s the tool that’s had the most work put into that problem specifically.

ChatGPT can generate images too, and it’s improved at text compared to earlier versions, but it still misspells things. Check every single word it gives you. Same goes for Gemini’s image features. Don’t trust any of them blind.

Adobe Express is another one worth a look, similar territory to Canva, template-driven, good for exporting at print sizes.

To get going: sign up for a free Canva account at canva.com, or if you want to try the AI-image route, sign up for ChatGPT at chatgpt.com or Ideogram at ideogram.ai. All have free tiers to start with, though limits vary and change, so check what’s on offer when you sign up rather than trusting what I say here.

Try it yourself

Here are three prompts, each doing something different. Copy them, paste them into whichever tool you’re using, then adjust.

First, use this to get layout and content ideas before you touch any image tool. This works well in ChatGPT, Claude, or Gemini, just as plain text:

I'm making a poster for [event name], happening on [date] at [venue/location].
Suggest a layout structure using hierarchy: one dominant headline, one
medium-sized line for date/time/location, and small text for extra details
like ticket links or sponsors. Give me three options for wording the
headline, keeping it under 6 words each.

Second, use this in an image generator (Ideogram, ChatGPT, or Gemini) to get the artwork itself, treating any text it produces as rough and likely wrong:

Create a poster background image for a [type of event, e.g. summer music
festival]. Style: [bold and colourful / minimalist and elegant / vintage
poster style]. Leave clear open space in the upper third for a headline and
in the lower third for event details. No readable text needed, just the
visual mood and composition.

Third, once you’ve got your artwork, use this in Canva’s Magic Design or similar to help you structure the actual editable text layer:

I have a poster background image. I need to add: a headline saying
"[your event name]", a subheading with the date and venue, and small
text listing three sponsor names. Suggest font pairings and sizes that
keep clear visual hierarchy, suitable for printing at A3 size.

Notice what’s happening there. The first prompt is thinking. The second is mood and image. The third is actual construction, where you type the real words yourself so they’re spelled how you meant them. That middle step, handing text duties to a design tool rather than the image model, is the one thing that saves you the most grief.

Check the result

Once you’ve got something that looks like a poster, don’t just admire it. Test it.

Read every single word out loud. Not skim it, read it, slowly, especially the date and the venue name. This is the single most common AI poster failure: wrong dates. An AI has no idea what day of the week the 14th of June actually falls on, or whether your venue is spelled “Odeon” or “Odean”. Proofread like your event depends on it, because it does.

Check the hierarchy actually works. Stand back from your screen, or better, shrink the image down small on your phone, the size someone might see it scrolling past on social media. Can you tell what the event is and when it’s happening in about two seconds? If you can’t, the hierarchy’s off and you need to make the headline bigger or the details smaller.

If you’re printing, export at the right size and resolution. For a physical poster that’s usually A4 or A3, at 300 dpi. Most design tools have a “print” export option that handles this for you, just make sure you pick it rather than exporting a low-res web version by accident.

And check it on a phone screen too, even if you’re printing it, because someone’s going to photograph it and post it, and it needs to survive that.

If it doesn’t work

A few things worth knowing before you get frustrated.

Don’t hand an AI tool a photo of a real person’s face and ask it to put them on a poster, especially not without their permission. That’s a genuine privacy and consent issue, not just a technical one. Same goes for using someone’s likeness generated by AI, it’s still their likeness.

Don’t ask an AI to recreate a brand logo, whether that’s a football club crest, a company logo, or anything trademarked. Aside from AI generally being poor at reproducing logos accurately anyway, you can run into real legal trouble using someone else’s protected branding on a poster you’re distributing, even informally.

If the text keeps coming out wrong no matter how many times you regenerate, stop fighting it. That’s your signal to pivot to a design tool. Generate the background image once you’re happy with the mood, then open it in Canva or Adobe Express and type the words yourself. It sounds like admitting defeat but it’s actually just using each tool for what it’s good at.

Expect misspellings especially with unusual words, place names, or anything with numbers mixed into text, like “14th” or postcodes. These are the areas AI text-in-image still trips over most.

Keep exploring

If you want to go further with AI-generated visuals more broadly, have a look at How to get AI to create an image. If your poster needs a logo and you don’t have one yet, there’s How to get AI to make a logo. And if you’re pulling together a pitch or a slideshow to go alongside your event, How to get AI to make a presentation covers that ground.

Sources and review notes

Review date: 15 September 2026 — every source above was opened and checked against the text on that date.