I upgraded my AI image prompts using Gemini’s advice — specifying lighting, layouts, and even fonts changed everything
A few small tweaks to your prompts can go a long way with AI images
Sign up for breaking news, reviews, opinion, top tech deals, and more.
You are now subscribed
Your newsletter sign-up was successful
Join the club
Get full access to premium articles, exclusive features and a growing list of member rewards.
It wasn’t until I saw a tweet from Gemini on X today that I realized I’d been approaching AI image prompts all wrong.
I’ve been using Gemini, well, really its default image generator, Nano Banana, instead of ChatGPT for images ever since it launched. In my experience, it tends to get closer to what I want on the first try. That’s not to say ChatGPT is bad at images, but I often find myself refining prompts more to get there.
Still, I hadn’t really questioned how I was prompting. I’d been sticking to simple descriptions. The problem is that “decent” is about as far as that approach takes you.
Article continues belowGemini’s tweet made a simple point: if you want better images, you need to be more specific. So, let’s unpack what that actually looks like.
Get the most out of your Nano Banana generations by establishing the story, subject, and style.Try including...Subject: Who or what is in the image? (Ex: A fluffy calico cat)Composition: How is the shot framed? (Ex: Extreme close-up)Action: What is happening? (Ex: Brewing a…April 13, 2026
From simple prompts to creative direction
Most of us start with the basics: subject, style, action, maybe a location.
That gets you something like:
“A fluffy orange cat playing with a ball in a watercolor style.”
Perfectly fine, but also pretty generic.
Sign up for breaking news, reviews, opinion, top tech deals, and more.
Gemini’s suggestion is to go further by layering in more specific instructions. The full list includes composition and aspect ratio, camera and lighting details, text integration, factual constraints, and reference inputs.
You don’t need all of them every time. But adding just one or two can make a noticeable difference. Here are the three that had the biggest impact for me.
Define the frame with composition and aspect ratio
This is one of the simplest upgrades, and one of the most effective. If you don’t specify an aspect ratio, you’re leaving the framing entirely up to the AI. That’s why images can feel oddly cropped or just slightly “off”.
Adding something like:
- 16:9 for landscape
- 4:5 for portrait (a great all-rounder)
- 3:4 or 2:3 for variation
…immediately makes the output feel more intentional. You can also use shorthand like --ar 16:9 to save time. And if you're making the image for a poster, then tell Gemini it is going to be a poster, and it will help make a better image.
Think like a cinematographer with lighting and camera details
This is where things start to look dramatically better. Instead of just describing the subject, you’re directing how it’s captured. Lighting, in particular, has a huge impact on mood and realism. For example:
“Golden hour backlighting creating long shadows”
“Rembrandt lighting with deep shadow contrast”
Here’s a full example:
Prompt:
A close-up portrait of an elderly man with weathered skin, Rembrandt lighting with a deep chiaroscuro effect, one side of the face in heavy shadow, sharp rim light catching the silver hair, 8k resolution, highly detailed textures, --ar 16:9
It’s the difference between a generated image and something that feels deliberately shot.
Control the text instead of hoping for the best
AI image tools are much better at rendering text than they used to be, but only if you’re explicit. Don’t just say “add text”. Tell it: what the text is, where it goes, and what it should look like.
Gemini’s example is:
“The headline ‘URBAN EXPLORER’ rendered in bold, white, sans-serif font at the top.”
Expanded into a full prompt it looks like:
Prompt:
A premium black cotton T-shirt flat-lay on a grey concrete background. The headline “URBAN EXPLORER” is rendered in a bold, white, sans-serif font centered on the chest. Soft top-down studio lighting with subtle shadows to show fabric texture. High resolution, minimalist aesthetic, --ar 16:9
(For something like this, a 4:5 ratio would usually work better—but 16:9 fits web layouts nicely.)
Experiment with these three aspects and you’ll find that your images start to dramatically improve. And if you need help with a specific prompt, don’t hesitate to ask Gemini to suggest one, or show you how to improve one you’ve written, after all, that’s what it's there for!
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!
And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.

➡️ Read our full guide to the best business laptops
1. Best overall:
Dell Precision 5690
2. Best on a budget:
Acer Aspire 5
3. Best MacBook:
Apple MacBook Pro 14-inch (M4)

Graham is the Senior Editor for AI at TechRadar. With over 25 years of experience in both online and print journalism, Graham has worked for various market-leading tech brands including Computeractive, PC Pro, iMore, MacFormat, Mac|Life, Maximum PC, and more. He specializes in reporting on everything to do with AI and has appeared on BBC TV shows like BBC One Breakfast and on Radio 4 commenting on the latest trends in tech. Graham has an honors degree in Computer Science and spends his spare time podcasting and blogging.
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.