Write Prompts That Get You the Right Picture

← Back to Issue #10

Go Deeper Guide  •  Companion to Issue #10

Write Prompts That Get You the Right Picture

This issue's prompt got you your first Firefly image. This guide shows you how that prompt is actually built — so instead of hoping you get lucky, you can describe exactly the picture you want, and get it in one or two tries.

Why This Is Worth Ten Minutes

The single biggest reason people give up on AI image tools after one try: they type something vague like "a picture of a family" and get back something generic, or off, or just not what they pictured. That's not a Firefly problem — it's a description problem. The fix is learning the shape of a good prompt, and once you see it, you'll use it every time.

Before You Start — Three Things to Know

• It's not one-and-done — and that's fine. Even with a great prompt, you'll usually generate a few times before one feels right. Every generation is free on Firefly's free plan (within your daily limit), so there's no cost to trying again.
• It's for pictures that don't exist yet. Firefly makes illustrations, concepts, and photorealistic scenes — not a real photo of a real, specific person or place you already know. Don't use it to fake a photo of an actual event.
• Hands and text are still the weak spots. AI image tools, Firefly included, sometimes render hands strangely or garble any text you ask it to include in the image. Avoid asking for readable text in the image itself — add that separately in whatever program you're using to build your flyer or card.

Step 1: Open Firefly and Find the Text Box

Getting there

Go to firefly.adobe.com and sign in with a free Adobe account, or use your existing Google account to sign in instead — no new password to remember. You'll land on the "Text to Image" tool automatically, which is just a big text box waiting for your description.

Step 2: Build Your Prompt in Three Parts

The formula

Every prompt that reliably gets a good result has three parts. Here's this issue's homepage prompt, broken down:

1. The Subject — who or what is in the picture
   “A retired couple sitting on a porch together, laughing”
2. The Setting — where it's happening, and any specific detail
   “A glass of iced tea on the table between them”
3. The Style — the look and mood you're after
   “Warm, sunlit, golden afternoon light, photorealistic style”

Subject, Setting, Style. Leave any one out and Firefly has to guess — which is usually where a vague result comes from. Include all three and you'll narrow in fast.

“A warm, sunlit photo of a retired couple sitting on a porch together, laughing, a glass of iced tea on the table between them, golden afternoon light, photorealistic style.”

Step 3: Generate, Compare, Refine

Getting to the right one

Click Generate and Firefly hands you four versions at once. Look them over — then either pick one, or refine your prompt and try again. A few refining moves that work well:

• Too generic? Add a specific detail — an object, a color, a season.
• Wrong mood? Add a lighting or time-of-day cue — "soft morning light," "warm evening glow."
• Wrong style entirely? Firefly has a Styles panel along the side — try "Photo" for realism or "Art" for something more illustrated, like a card or flyer might call for.

Once you have one you like, click it, then Download. It saves as a standard image file you can drop into anything — a Word document, a Canva flyer, an email, a printed card.

Try the Formula on Something Else

Same three-part shape, aimed at two other everyday uses.

A birthday card for a grandchild

“A cheerful cartoon-style illustration of a birthday cake with seven candles, colorful balloons in the background, bright and playful, children's book art style.”

A title slide for a talk or presentation

“A wide, calm image of a lighthouse on a rocky coastline at sunrise, soft pastel colors, minimalist and clean, plenty of empty sky for text to sit on top of.”

Notice that last one — "plenty of empty sky for text to sit on top of" is a Setting detail that solves a real, practical problem: leaving room for a title once you drop the image into your slide.

What Firefly Still Isn't Great At

• Readable text inside the image — it'll come out garbled more often than not. Add text afterward in your card, flyer, or slide program.
• An exact likeness of a real, specific person. It's for scenes and concepts, not portraits of people it's never seen.
• Hands, in close detail, occasionally look off. If a generated image has a person with visible hands, look closely before you use it.

Your Turn — Try It This Week

• Open Firefly and generate an image using the Subject / Setting / Style formula for something you actually need — a card, a flyer, a slide.

• If the first result isn't right, refine one part of the prompt and try again rather than starting over.

• Download an image you like and actually use it somewhere this week.

One Last Thing

The goal here isn't to become a prompt expert. It's to stop settling for "close enough" from a stock photo site when what you actually pictured is one clear description away.

Talk soon,
Gary
Founder, AI Unlocked

← Back to Issue #10

AI Unlocked  •  aiunlocked.us  •  Your trusted guide on the AI journey

Helping adults 55+ navigate AI with confidence — no hype, no jargon.