Banana Prompts to Copy, Paste, and Post

Banana prompts are the ready-made instructions people paste into Nano Banana, the nickname for Google's Gemini image model (officially Gemini 2.5 Flash Image), and this page is a roundup of them grouped by the kind of result you want. The grid up top is the living collection, already written and paired with sample images. Below I have sorted the ideas into clusters so you can scan to the look you are after, grab the wording, and run it on your own photo instead of starting from a blank box.

How these banana prompts are sorted

Rather than dump a hundred random lines on you, I bucketed them by intent. That is usually how people actually search: someone wants a couple edit, or a film look, or a tiny figurine of themselves, not just any prompt. Each cluster below covers what it is best at, the kind of source photo that works, and a short wording you can lift straight away. The full set lives in the grid, but these notes tell you which corner of it to dig into.

• Solo portraits and selfies, the biggest pile by far.
• Couple and group shots, for two or more faces.
• Retro and film-era looks, from 90s prints to old passport photos.
• Toy, figurine, and 3D versions of a person.
• Festival, seasonal, and themed edits tied to a date or mood.

Solo portrait and selfie banana prompts

This is where most of the collection sits, because it is the easiest win. You upload one clear selfie and the model restyles everything around your face while keeping you recognizable. The popular ones right now lean editorial and filmic rather than cartoonish. Think soft magazine lighting, muted color, sharp eyes. Studio headshots for profile pictures are a steady favorite too, since people use them on LinkedIn and dating apps.

What to start from

A front-facing photo, good light, no heavy filter already baked in. The cleaner the input, the less the model has to guess. If you feed it a dim, blurry shot, no prompt in this collection will save it.

Wording to try

Something like: a portrait of me, keep my exact face and skin tone, soft editorial lighting, muted warm color, sharp focus on the eyes, plain backdrop. Short and specific beats a paragraph of adjectives.

Couple and group banana prompts

The couple prompts in this collection are written to hold two faces at once, which is harder than it sounds. The trick baked into them is naming both people and telling the model to keep both faces, then setting one shared scene so the lighting matches. These get shared a lot for anniversaries and for turning a phone snap into something that looks staged on purpose.

Group shots of three or more are the least reliable corner of the whole set. The more faces you ask it to preserve, the more one of them drifts. If you have a group photo, expect to run it a few extra times, or crop down to two people for a cleaner result.

Retro and film-era banana prompts

Vintage is having a long moment, so this cluster is deep. The prompts here recreate specific eras instead of a vague old feel, and that specificity is what makes them land. A 90s film print reads differently from a 70s Polaroid or a faded 2000s digicam shot, and the collection has wording for each.

Tip: name the decade and the film stock, not just the word vintage. '35mm film, 1990s, soft grain, faded warm color' gives the model something concrete to aim at, where 'old photo' leaves it guessing.

These looks forgive a slightly imperfect source photo because grain and color shifts hide minor flaws. That makes the retro cluster a good place to start if your only photo is a bit rough.

Toy, figurine, and 3D banana prompts

The boxed collectible figurine of yourself refuses to die, and honestly it earns its spot because it reads instantly and it is fun to send. This corner of the collection turns a person into a small plastic toy on a desk, a boxed action figure, or a clay-style 3D character. The wording usually pins the scale and the material so the model commits to the toy look instead of half-doing it.

Here is a quick map of the main clusters so you can jump to the right one.

Festival and seasonal banana prompts

This cluster moves with the calendar. Around big festivals the collection fills up with themed edits, the kind where you keep your face but the scene turns into a celebration. These are popular because they are timely and easy to share the same day. The wording tends to name the festival, the colors tied to it, and the mood, then locks your face so the result is still you in the scene rather than a stranger.

Seasonal looks work the same way. A warm autumn portrait, a snowy winter scene, a bright summer beach edit. Each is just a portrait prompt with the setting swapped, which is why they are easy to adapt from any of the other clusters above.

How to adapt any banana prompt in this collection

Every prompt here is a starting point, not a rule. The fastest way to make one yours is to keep its skeleton and change one part at a time.

1. Copy a line from the cluster closest to what you want.
2. Open Gemini and upload one sharp, well-lit photo.
3. Add 'keep the same face and skin tone' if a person needs to stay recognizable.
4. Run it, then change a single detail, like the light or the background, and run again.
5. Save the exact wording that worked so you can reuse the look next time.

Results wobble between runs, and that actually helps you. The second or third attempt is often the keeper, so do not judge a prompt on its first output.

Picking the right banana prompts for your photo

If you are not sure where to start, let the photo decide. A clean front-facing selfie can handle almost anything in the portrait or figurine clusters. A slightly rough photo does better in the retro looks, where grain hides flaws. A shot with two people belongs in the couple cluster, and a group is best cropped down before you try. Match the prompt to the input you already have and you will spend less time fighting the model.

One note on cost, since people ask. Copying anything from this collection is free. The Gemini app has a free tier that covers everyday edits, and heavy or batch use runs through the paid API at a few cents an image. Try any prompt free first, and only think about paying if you are generating in bulk. Pick a cluster, grab a line, drop in your photo, and run it three times before you decide whether it is the one.