Prompts written for Google Gemini, including the photo-editing style prompts that take an image you upload and return an edited version. Copy any one of them and paste it into Gemini as-is.
Black and white fine art style portrait of a young woman, with long, slightly messy hair framing her delicate face, smoo...
Create a digital image of a woman wearing a bright, vibrant saree that has a geometric checkered pattern in shades of pi...
Strictly preserve the original face identity and pose. Replace the outfit with a haute couture bespoke suit (Color: Blac...
Strictly preserve the original face identity and pose. Replace the outfit with a haute couture bespoke suit (Color: Blac...
Keep the facial features of the person in the uploaded image exactly consistent . Style : A cinematic, emotional portrai...
Photorealistic cinematic portrait, same face, sitting on old stone steps in a quiet historic alley in early morning soft...
A man in an elegant tuxedo holding a glass of champagne, sitting casually on a luxury rooftop lounge with city skyscrape...
A man in trendy streetwear, oversized hoodie, ripped jeans, and sneakers, leaning casually against a graffiti wall with ...
Explore the full library — 1,000+ prompts, all free to copy.
Go to gemini.google.com and start a new chat.
Editing prompts need a source image. Attach it before sending the prompt.
Copy any prompt below and send it unchanged the first time.
Gemini keeps the conversation, so you can say "make the background darker" instead of rewriting the whole prompt.
Gemini is currently the most reliable of the mainstream models at two things: rendering readable text inside an image, and editing a photo you give it rather than inventing one from nothing. That makes it the right choice for posters, packaging mockups, UI screens, infographics and anything where a word has to come out spelled correctly. Midjourney will usually produce the more striking picture; Gemini will produce the one where the label is legible.
A generation prompt describes an image that does not exist yet. An editing prompt describes a change to one you upload — restyle this portrait, change the background, show this outfit in three colourways. Editing prompts need to name what to keep as clearly as what to change, otherwise the model redraws the whole frame. The editing prompts in this collection are written that way, which is why the face stays recognisable.
Put the exact words you want in quotation marks and state where they should sit in the frame. Keep it to a few words: accuracy falls off quickly past a short headline. If a letter comes out wrong, regenerate before rewriting — text rendering varies run to run, and the prompt is often already correct.
The prompts here are free to copy. Gemini itself has a free tier that covers image generation, with higher limits on paid Google AI plans.
Yes. Upload the image first, then send a prompt describing the change. Say explicitly what should stay the same — the face, the pose, the framing — or the model may redraw more than you intended.
Text rendering is sampled like everything else, so it varies between runs. Put the exact wording in quotes, keep it short, and regenerate a few times before changing the prompt.
The descriptive part does. Midjourney parameter flags such as --ar or --v do not — Gemini treats them as literal text, so remove them and describe the aspect ratio in words instead.