Gemini Polaroid Prompt: Copy-Paste Prompts That Work
September 14, 2026
A Gemini Polaroid prompt is a short instruction you give Google Gemini, along with one or more of your own photos, asking it to turn them into a single retro instant photo: soft flash, faded color, film grain, and that white frame with the wider strip at the bottom. The trend's two classic versions are you standing next to someone famous, and you hugging your younger self.
Here's the one to start with. Upload a clear photo of yourself, then paste:
Turn this photo into a vintage Polaroid instant photo. Keep my face, hair, and expression exactly the same. Add a white Polaroid frame with a wider bottom border, a slightly overexposed on-camera flash, soft warm faded colors, light film grain, and a plain off-white curtain behind me.
Below: prompts for the other versions, why faces drift, and the watermark and consent questions to settle before you post.
How to Upload Photos to Gemini for the Polaroid Look
The trend runs on Gemini's image editing, which Google calls Nano Banana. The first Nano Banana model arrived in the Gemini app in August 2025, and as of September 2026 Google's help pages name Nano Banana 2, which Google also calls Gemini 3.1 Flash Image, as the model behind image creation and editing in Gemini. Google says it keeps the resemblance of up to five characters in one workflow. If your Gemini model is set to Flash-Lite, Gemini uses Nano Banana 2 Lite instead, which isn't designed for multiple reference photos — switch to Flash or Pro for the prompts below that use two photos.
The steps are the same on a phone or a computer:
- Open a new chat in Gemini.
- Use the add-file button in the prompt bar to attach your photos. You can attach more than one, and Gemini will combine them into a new image.
- Paste your prompt and send.
- Refine in the same chat, one change at a time.
Two limits from Google's image help article are worth knowing up front. Editing uploaded images isn't available to users under 18. And there are daily limits on how many images you can make, so don't burn attempts on blurry source photos.
If you're doing this on your phone and the app itself is new to you, our guide to using AI on your phone covers the basics.
What makes a good source photo:
| Photo quality | What happens | Fix |
|---|---|---|
| Face small in the frame, or a group shot | Gemini has few details to work from, so the face drifts | Crop so the face fills a good part of the image |
| Side angle or sunglasses | Eyes and jawline get guessed | Use a front-facing photo with eyes visible |
| Dark or motion-blurred | Features get invented to fill the gaps | Pick a photo in daylight or even indoor light |
Copy-Paste Gemini AI Polaroid Prompts
Each prompt below is written for one job. Swap the bracketed parts for your own details.
Hugging your younger self
Upload two photos: a current one of you, and a childhood photo. Say which is which in the prompt, because Gemini doesn't know the order you meant.
Using the two photos I uploaded, create one vintage Polaroid photo. The first photo is me now. The second photo is me as a child. Show my adult self gently hugging my younger self from behind, both smiling at the camera. Keep both faces true to the original photos. Plain off-white curtain background, soft on-camera flash, warm faded colors, light film grain, white Polaroid frame with a wider bottom border.
Family or friends group photo
Upload one photo per person, or one group photo where everyone's face is clear. Name each person by position.
Combine these photos into one Polaroid-style group photo taken at a [birthday party / kitchen table / backyard] in the 1990s. From left to right: [person from photo 1], [person from photo 2], [person from photo 3]. Everyone keeps their own face, hair, and skin tone from their photo. Arms around each other, relaxed and laughing. Direct flash, slightly soft focus, faded warm colors, visible film grain, white Polaroid frame.
Keep it small. The more people you add, the more likely two faces start borrowing features from each other.
With a public figure, kept obviously playful
This version needs the most care. Read the consent section below first.
Create a playful, clearly staged Polaroid-style photo of me standing next to a cardboard cutout of [public figure]. Keep my face exactly as in my photo. Party setting, bright on-camera flash, faded colors, film grain, white Polaroid frame. Handwritten caption on the bottom border: "AI edit".
If you want more scene ideas to rotate through, a random prompt generator can hand you settings and props you wouldn't have thought of.
Why Faces Change, and Prompts That Fix It
Gemini isn't pasting your face onto a template. It's generating a new image that uses your photo as a reference. The whole image is regenerated, including your face. That's why the lighting looks so convincing, and also why your nose can come back slightly different, or your younger self can look like a cousin.
The fixes are specific. Send them as follow-up messages in the same chat:
| Problem | Follow-up prompt |
|---|---|
| Face drifted | Keep everything else, but make my face match my original photo exactly: same eye shape, nose, jawline, and eyebrows. |
| Wrong frame (square crop, black border, no frame) | Put the photo inside a classic white Polaroid instant film frame, with the thicker white border at the bottom. |
| Too clean, looks like a phone photo | Make it look like a real instant photo from the 1990s: harsh on-camera flash, slightly washed-out colors, soft focus at the edges, visible grain, a tiny bit of blur. |
| Two people's faces blended | The person on the left must look only like photo 1, and the person on the right only like photo 2. Don't mix their features. |
| Background too busy | Replace the background with a plain off-white curtain. Don't change the people. |
If a chat has gone many rounds and the face keeps getting worse, start a new chat with the original photos. Errors compound when you edit an edit of an edit.
For the general version of this skill, how to describe style, lighting, and framing for any image, see how to write a prompt for an AI image. If you've found a Polaroid edit you like and want to work out what produced it, our look at free image-to-prompt tools covers what they get right and wrong.
Watermarks on Gemini Polaroid Images
Every image Gemini creates or edits carries two built-in AI markers, and usually a third, visible one:
- SynthID. An invisible watermark from Google DeepMind's SynthID. It's embedded in the image itself. You can upload an image to Gemini and ask whether it was made or edited with Google AI.
- Content Credentials. C2PA metadata describing where the image came from.
- A visible watermark. Google's watermark settings page says you can manage the visible mark in settings, though in India, South Korea, and Vietnam the option only appears with an AI Ultra subscription.
Turning off the visible mark doesn't remove the other two, and SynthID survives ordinary edits — DeepMind says it holds up through filters, color changes, and lossy compression. But don't treat either marker as a guarantee. Google is explicit that SynthID isn't foolproof against extreme image manipulation, and that C2PA metadata can be stripped by hand or simply lost when a file is edited. Detection also comes back as a confidence level, not a yes or no.
The practical version: cropping the corner off a Gemini Polaroid probably doesn't make it untraceable, but "probably" is the strongest claim available. Disclose the edit yourself rather than relying on automatic detection to do it for you.
Privacy and Consent: Rules Before You Post
Only use photos of people who agree to it. A friend, a sibling, or a parent should say yes before their face goes into an upload, and again before the result goes online. For kids' photos, the parent decides.
Think twice about real celebrities. Google's Generative AI Prohibited Use Policy bars impersonating a person, living or dead, without explicit disclosure in order to deceive. A realistic photo of you with an actor you've never met gets screenshotted and passed around as real. If you make one, keep it obviously staged, label it as an AI edit, and don't use it to suggest a friendship, endorsement, or meeting. If Gemini declines a request, don't try to word your way around it.
Never pass it off as a real photo. "Look who I ran into" with no disclosure is where a harmless edit turns into misinformation. Say it's AI.
A Polaroid prompt that finally gives you the look you want is worth keeping, along with the follow-up fixes that got you there. Taku is an AI-native desktop workspace where you can mirror a creative setup someone has already tuned, run it, remix it for your own photos, and keep it for next time instead of rebuilding the prompt from memory. Taku is in Beta, and the Mac app is available now.
Key Points
- A Gemini Polaroid prompt pairs your uploaded photos with a request for a vintage instant-photo look: flash, faded color, grain, and a white frame with a wider bottom border.
- As of September 2026, Google names Nano Banana 2 (Gemini 3.1 Flash Image) as the model behind image editing in Gemini. Editing uploads requires being 18 or older.
- Faces change because Gemini redraws the whole image. Use clear, front-facing source photos and fix drift with one targeted follow-up at a time.
- Gemini images carry an invisible SynthID watermark and Content Credentials, whether or not the visible mark is shown. Neither is foolproof against heavy editing, so disclose AI edits yourself.
- Get consent from everyone pictured, label AI edits, and don't post realistic celebrity images as if they were real.
FAQ
What is the best Gemini Polaroid prompt?
The one that names the look and protects your face: vintage Polaroid, keep my face exactly the same, white frame with a wider bottom border, on-camera flash, faded colors, film grain. Then adjust one detail at a time.
How do I use a Google Gemini Polaroid prompt with two photos?
Attach both photos in the same message, then say in the prompt which photo is which, such as "the first photo is me now, the second is me as a child." Without that, Gemini has to guess who's who and may mix features.
Why does my face look different in the Gemini AI Polaroid image?
Gemini generates a new image using your photo as a reference rather than editing pixels in place, so your features get redrawn. Sharp, front-facing source photos reduce the drift.
Can I remove the watermark from a Gemini Polaroid photo?
Depending on your country and plan, Google lets some users turn off the visible watermark in Gemini's settings. The invisible SynthID watermark and the Content Credentials metadata stay either way, and SynthID is designed to survive common edits like filters, color changes, and compression. It isn't guaranteed, though: Google says SynthID isn't foolproof against extreme manipulation, and C2PA metadata can be removed or lost when a file is edited. Treat the markers as a backstop, not a substitute for saying the image is AI.
Is it okay to make a Gemini Polaroid with a celebrity?
It's risky. Google's prohibited use policy bars impersonating a real person without disclosure to deceive, and realistic celebrity photos get shared as if they were real. If you make one, keep it clearly staged and label it as AI.
Does the Gemini Polaroid prompt work on a phone?
Yes. The same prompts work in the Gemini mobile app. Attach photos from your camera roll, paste the prompt, and refine in the same chat.