Free ChatGPT Image Prompts for GPT-4o

Design a bold portrait event poster. The big headline reads [MOVIE NIGHT IN THE PARK] in a heavy condensed sans-serif. Under it, smaller text reads [Friday, August 9 - 7pm - Free admission]. Warm sunset palette, strong contrast so every word is crisp and legible.

GPT-4o renders exact quoted words far better than older models, so a poster with real headlines and dates comes out sharp instead of gibberish.

Create a clean logo for a coffee brand called [EMBER & OAK]. Show the full name as a wordmark in a modern serif, with a simple flame-and-leaf mark above it. Flat two-color vector, centered on white, no extra text or letters.

Naming the exact brand words and asking for a wordmark gets real, readable letterforms instead of the fake text older image models produce.

Make a meme image. Top caption reads [ME AT 3PM] and bottom caption reads [ME AFTER ONE COFFEE]. Center image is a [sleepy cartoon cat on the left, wide-eyed cartoon cat on the right], split down the middle. Bold white meme font, black outline, spelled correctly.

Quoting each caption line keeps the spelling right, which is exactly where normal meme generators fall apart.

Using the photo I just uploaded, keep the person exactly the same and replace the background with [a sunny beach at golden hour]. Match the light and color on the subject to the new scene so it reads as one real photo.

GPT-4o edits an image you upload from a plain-English instruction, so you describe the one change instead of regenerating from scratch.

Take the image I uploaded and remove the [trash can on the left]. Fill that space naturally using the surrounding [brick wall]. Do not change anything else in the photo.

Telling it precisely what to remove and what to fill with gives a clean edit while the rest of the shot stays untouched.

From the photo I uploaded, keep the same room and furniture but change the view outside the window to [winter with falling snow], and warm the indoor lighting to a cozy evening glow.

The keep-X-change-Y pattern works because the model preserves every part of the picture you did not mention.

Create a vertical infographic titled [5 Steps to Better Sleep]. Five numbered rows, each a short label with a simple icon: [1 Set a bedtime], [2 No screens], [3 Cool room], [4 No caffeine], [5 Wind down]. Calm blue palette, lots of white space, all text spelled correctly.

GPT-4o is one of the few image tools that keeps infographic labels readable, so you get a shareable graphic in a single pass.

Draw a simple left-to-right flow diagram of [how a coffee order moves from app to pickup]. Four labeled boxes joined by arrows: [Order placed] to [Sent to barista] to [Drink made] to [Ready for pickup]. Flat, minimal, clear labels.

Naming each box and the arrow order gives a diagram that maps your real process, not a random-looking chart.

Show a [smartphone] held in one hand with a [meditation app] home screen on it. The screen shows a title [Breathe], a large round Start button, and three sessions listed: [Morning Calm], [Focus], [Deep Sleep]. Realistic photo of the phone, crisp UI on screen.

Putting real UI labels on a realistic device gives a believable product mockup for a pitch deck or store page.

A photorealistic portrait of a [55-year-old fisherman] with a weathered face, looking into the camera, soft window light from the left, 85mm lens, shallow depth of field, natural skin texture, no text.

GPT-4o handles real faces and directional light well, so a clear lens-and-light description gets a usable portrait fast.

A cinematic wide shot of [a lone diner glowing on an empty highway at night]. A neon sign reads [OPEN], wet asphalt reflects the lights, moody blue-and-red palette, subtle film grain, atmospheric.

A named subject plus mood, palette and a real readable sign keeps the scene both cinematic and coherent.

Studio product photo of a [glass bottle of cold brew] on a [polished stone slab], three-point lighting, soft reflections, condensation on the glass, deep shadow, catalog quality, plain background, no text.

The studio recipe gives clean, commercial-looking product shots you can drop straight onto a store page.

Create a [friendly robot mascot named Bolt: round white body, one big blue eye, small antenna]. Remember this exact design. First image: Bolt waving hello. I will then ask for Bolt in new poses and scenes, and he must stay the same character every time.

GPT-4o keeps a character consistent across a conversation, so you lock the design once and reuse it in later prompts.

Using the photo I uploaded, redraw the same scene as [a soft watercolor painting]. Keep the composition and the main shapes, but give it loose brushwork and gentle washes of color.

You can hand it a real photo and ask for a specific art style, and it keeps your composition while changing the look.

From the selfie I uploaded, make a [1990s anime cel] version of me. Keep my hairstyle, glasses and outfit recognizable, add hand-drawn line art, flat cel shading, and a soft city-at-dusk background.

Listing the features to keep, like hair and glasses, means you stay recognizable even after a big style change.