Gemini AI photo prompt guide: templates for better image edits
A practical guide to writing Gemini AI photo prompts: a simple structure, ready-to-use templates for portraits and products, and tips for editing your own photos.
On this page
A good Gemini AI photo prompt is specific about four things: the subject, the setting, the style and the light. Google's Gemini app can generate new images and edit photos you upload, using its Nano Banana image models. The difference between a generic result and a usable one is almost always the prompt.
This guide gives you a simple prompt formula, copy-ready templates for common jobs like portraits, product shots and photo edits, and the limits you should know before using results commercially.
The short answer
Start your Gemini prompt with an action verb (create, generate, edit), then describe the subject, what it is doing, the setting, the style (for example photorealistic), the lighting and the framing. When editing an uploaded photo, ask for one change at a time and say what must stay the same. Refine with short follow-up prompts instead of rewriting everything.
How Gemini handles photos
In the Gemini app you can type a prompt to generate an image, or upload your own photo and ask for edits. According to Google's Gemini help page, the app uses Nano Banana models: Nano Banana 2 for most image work, with Nano Banana Pro available to paid subscribers for more detailed redos. Paid plans download images at 2K resolution and free users at 1K. You must be 18 or over to edit images with a personal account.
Gemini images carry Google's invisible SynthID watermark and content credentials metadata, which mark them as AI-generated. Google has also shown a visible watermark on images, which became optional for some users during 2026. Features and limits change frequently, so check the help page for current details.
The Gemini AI photo prompt formula
Use this structure as a checklist:
| Element | What to specify | Example |
|---|---|---|
| Action | Create, generate, edit, replace, remove | "Create a photorealistic image of..." |
| Subject | Who or what, with key details | "a ceramic coffee mug in matte sage green" |
| Action or pose | What the subject is doing | "with steam rising from it" |
| Setting | Location, surface, background | "on a light oak table beside a window" |
| Light | Direction, quality, time of day | "soft morning side light" |
| Camera and framing | Angle, lens feel, depth of field | "close-up, shallow depth of field" |
| Style and use | Photo style, mood, aspect, purpose | "clean lifestyle style for a website banner" |
Photography vocabulary helps a lot. Words like "backlit," "golden hour," "overhead flat lay" or "85mm portrait look" give the model concrete targets. If any of these terms are unfamiliar, our photography terms guide is a quick reference.
Copy-ready Gemini prompt templates
Professional headshot style
Create a photorealistic professional headshot of a [age] [person description] wearing a [clothing], neutral gray studio background, soft key light from the left with gentle fill, shoulders-up framing, shallow depth of field, natural skin texture.
Product lifestyle image
Generate a photorealistic lifestyle photo of [product] on [surface] in a [room type], soft window light from the right, a few props such as [props], muted earthy color palette, leave empty space on the left for text.
Editing an uploaded photo
Edit this photo: replace the overcast sky with a clear late-afternoon sky. Keep the building, people and foreground exactly the same, and match the warmer light on the walls.
Background change
Keep the person exactly as they are, including face, hair and clothing. Replace the background with a softly blurred modern office with warm lights.
Food photography
Create a photorealistic overhead flat lay of [dish] on a dark slate surface, with [ingredients] scattered around, soft diffused light from the top left, rich natural colors, a linen napkin and a fork at the edge of the frame.
Removing distractions
Edit this photo: remove the trash can and the parked car on the right. Fill the area naturally with the matching sidewalk and wall. Do not change anything else.
Restyling a photo
Turn this photo into a watercolor illustration with loose brushstrokes and a white paper background, keeping the composition the same.
Tips for better results
- One change at a time. Multi-part edit requests are more likely to alter things you wanted kept. Make sequential edits.
- Say what must not change. "Keep the face and expression identical" protects the most important parts.
- Describe, do not just name. "Cozy" means less than "warm lamp light, knitted blanket, rain on the window."
- Use follow-ups. "Make the light warmer" or "zoom out slightly" refines without starting over.
- Give a purpose. Mentioning "for a website hero banner" or "for an Instagram post" nudges composition.
- Check details at full size. Look closely at hands, text, logos and edges before using an image.
Iterating on a result
Treat the first image as a draft. Good follow-up prompts are short and specific: "lower the camera angle," "make the background less busy," "warmer white balance," or "same image, but in a vertical 4:5 crop." If the model drifts away from what you liked, go back to the earlier version and branch from there rather than stacking more corrections.
Common mistakes and limits
- Brand-critical images: AI can change product details, colors or labels. For e-commerce listings, real photography with retouching is safer and more honest.
- Text in images: results have improved, but check spelling and kerning, or add text later in a design tool.
- People and rights: do not edit photos of others in misleading ways, and respect copyright and privacy, as Google's own terms require.
- Disclosure: some platforms and ad networks require AI-generated content to be labeled.
For bigger edits, many designers combine Gemini with Photoshop: generate or edit a piece in Gemini, then composite, color-match and clean up by hand. Our guides to using Nano Banana in Photoshop and to AI photo generators cover that hybrid workflow and the wider tool landscape.
When to hand it to a professional
AI is excellent for ideas, mood boards, social content and quick fixes. For product catalogs, campaign visuals and anything where accuracy and consistency matter, a human retoucher still does the final mile: matching color across a set, fixing edges, keeping products true to life and producing print-ready files. That is the work our image editing and graphic design service specializes in, and it pairs well with AI-generated drafts.
Frequently asked questions
Can Gemini edit my own photos?
Yes. Upload a photo in the Gemini app and describe the change. Editing requires a personal account holder to be 18 or older, and availability varies by region and account type.
What is the best Gemini AI photo prompt for realistic images?
Ask for a "photorealistic" image and describe real photographic conditions: light direction, time of day, lens feel and depth of field. Concrete camera language produces more believable results.
Are Gemini images watermarked?
Yes. Gemini images include an invisible SynthID watermark. A visible watermark has also been used, with options that vary by account and region.
Can I use Gemini images commercially?
Review Google's current terms and the rules of the platform where you will publish. Avoid using AI images in ways that could mislead customers about a product.
Have AI drafts that need a professional finish? Our image editing and design team refines, composites and color-matches images so they are accurate, consistent and ready to publish.
More useful paths
Use these paths to continue with the closest practical topic, service, or decision guide.
Spotted something wrong? Report an error on this page. We correct on the page and say what changed.