DREAMMMERUpload a photo, say what should change, and Google's Gemini 3.1 Flash Image model edits it while aiming to keep your face recognisable; results vary most when the pose or expression changes. 13 credits per 1K image, about $0.29; new accounts start with free welcome credits, and you need no Google AI plan or API key. Every new image, including a revision, is priced before you send.
Original examples generated with this model. Open an example to see its prompt and output details.
Each pair is a real run on one uploaded photo of a generated person, with the words that were typed. Drag the handle to compare, or use the words on your own photo.
Typed: “Put me in front of the Eiffel Tower in Paris at golden hour. Keep my face, hair, expression and clothes exactly the same; make the light, shadows and perspective look like a real travel photo.”
Typed: “Turn this into a professional corporate headshot: a tailored navy suit with a white shirt, a clean modern office background with soft light. Keep my face, skin tone and hairstyle exactly the same.”
Typed: “Change only my hairstyle to short curly hair. Keep my face, clothes, pose and the yellow background exactly the same.”
Nano Banana 2 is Google's Gemini 3.1 Flash Image model: fast, good at following one instruction about one photo, and good at keeping the person recognisable. On DREAMMMER you use it in a chat: attach your photo, type the change, and refine the result in plain words. 1K images cost 13 credits, 2K 20 and 4K 29; a failed generation is refunded.
Say the one thing that should change and what must stay: 'change only the shirt to a grey hoodie, keep my face, pose and background'. Clothes and background edits hold the likeness best; hair and lighting well; big changes of pose or expression less reliably.
'Put me in front of the Eiffel Tower at golden hour, keep my face and clothes.' The model rebuilds the scene, light and shadows around the same person. Name the time of day and the light so the result reads like a real photo.
The model also renders readable headlines and can help keep a product consistent across scenes, as the gallery below shows. Write the exact words, their relative size and where the empty space should be, and check the lettering before you use it.
The jobs this model handles well on DREAMMMER, with how to brief it.
Attach a clear selfie and ask for a corporate headshot: a tailored suit, a clean office background, soft light, and 'keep my face, skin tone and hairstyle exactly the same'. The second example above was made exactly this way.
For a plain outfit swap you do not need a prompt: the Clothes Changer tool runs on Nano Banana 2.1, the newer edit model, for 7 credits a photo. Type the outfit, start, and keep adjusting in the chat.
Describe the place and the light, or use the one-click Background Changer (7 credits) for a studio wall, a street or a beach. The tool is designed to keep the person where they are, at the same size, and to match the lighting to the new scene; results vary with the original photo.
Run a faded or damaged print through the Photo Restore tool first, then bring the clean version here for the edit: a new background, a colour version, or the same person in a new setting.
Each revision uses the previous result as its reference. Compare what changes and what stays recognizable.
A joyful studio product photograph of a sculptural sunshine-yellow wireless speaker with a dark navy woven grille and small readable label PULSE. Rounded triangular silhouette, brushed silver volume wheel, navy geometric plinth against pale sky blue. Punchy but soft light, realistic fabric detail, modern approachable consumer design. Full composition, no border, no watermark.
Use the supplied image as the product reference. Preserve the exact product silhouette, construction, colors, logo and readable label. Replace only the setting with a sunlit sculptural stone terrace overlooking a calm blue lake and distant mountains. Keep a similar camera angle, realistic scale and the complete product in frame. Premium commercial photography, precise natural shadows and coherent reflections. No extra text, no watermark.
Use this result as the reference for one more edit. Preserve the exact product, label, shape, surface, lake, mountain background and camera composition. Change the daylight to a refined blue-hour mood with a warm amber rim light on the product and subtle warm reflections. Keep the important details legible and the product recognizable. No added objects or text, no watermark.
To reproduce an edit, attach the previous image as a reference along with the displayed prompt.
Tell the agent the subject, composition, and intended use. Add a reference when the selected model supports it, and say which details should be preserved.
This page starts with the featured model selected. Review the aspect ratio, available output size, and credit estimate. For video, also choose a duration that fits the action.
Open the completed work and inspect the details. Continue in chat with a specific revision, or download a result you want to use. Check visible text and listen to generated sound before publishing.
These are the settings connected in DREAMMMER. The table shows output credits without reference inputs; your composer estimate includes the inputs and settings of the actual request. Use your free welcome credits to try images at 1K. 2K and 4K require paid access; actual dimensions depend on model and aspect ratio.
| Output | Credits |
|---|---|
| 1K · per image | 13 |
| 2K · per image | 20 |
| 4K · per image | 29 |
Google offers Nano Banana 2 free in the Gemini app, with usage limits and 1K downloads. On DREAMMMER you need no Google AI plan or API key: new accounts get free welcome credits with no card, and one 1K image costs 13 credits, so the welcome credits cover your first edits. After that the smallest pack is $9 for 400 credits, about 30 images at 1K. A failed generation is refunded.
Nano Banana 2 is preselected in the composer at the top of this page. Attach your photo and type the change in plain words: 'put me in front of the Eiffel Tower, keep my face', 'change only the shirt to a grey hoodie'. The result opens in a chat where you can ask for the next change; each revision is a new image at the same price, and the composer shows the credit estimate before you send. For the most common edits, the one-click tools for clothes, background, object removal and restoration run on Nano Banana 2.1 with no prompt at all.
No. Nano Banana 2 is the name of Google's Gemini 3.1 Flash Image model, and Google's own places to use it are the Gemini app and Google AI Studio. DREAMMMER is an independent editor that runs the model through its API next to other image and video models, so you pay per image, need no plan, and can keep working on the result with the Agent.
Mostly, yes: keeping the person recognisable is one of the model's strengths, and the three edits above were made from one uploaded photo, but it cannot promise a pixel-perfect match. Changes of clothes and background hold best, hair and lighting well, big changes of pose or expression less so. Start from a clear, well-lit photo, say exactly what must not change, for example 'keep my eyes, nose, mouth, face shape, skin tone and proportions unchanged; only replace the background with a modern office', and check the result before you share it.
Yes, and one change at a time works best: 'change only the hairstyle to short curly hair, keep everything else'. Each edit is a new 1K image for 13 credits here. Nano Banana 2 powers this editor; the one-click Clothes Changer and Background Changer tools run on the newer Nano Banana 2.1 for 7 credits a photo.
Name the one change, name what stays (face, pose, background), and describe the light if you move the person somewhere, such as 'golden hour, natural shadows'. Short, specific sentences beat long lists. The 'Typed:' lines under the examples above are prompts that worked.
Both edit an uploaded photo from a text instruction. Nano Banana 2 is built for fast edits that keep the subject consistent, while GPT Image models are often picked for typography and layout. Both are available on DREAMMMER, so you can send the same photo to each, compare, and pay per image instead of per subscription.
1K (about 1024 px) for 13 credits, 2K for 20 and 4K for 29, chosen in the composer before you send; downloads are PNG. You own what you create and may use it under our Terms of Service, provided you have the right to use the photo. Downloads from an account that has never paid carry a small DREAMMMER mark in the corner; any credit pack or subscription removes it.
Uploads and results are private to your account by default and are not shown publicly when you share nothing. One exception, stated in our Privacy Policy: if an account has never made a purchase, selected work made with free credits may be featured in the public showcase after review, without your name or email; this never applies to an account that has made a purchase. To make an edit, the photo is sent to the provider that runs the model under confidentiality terms described in the Privacy Policy. You can access, export or delete your data from your account settings or by contacting us.
You can explore the page and prepare your prompt first. Sign in when you are ready to generate; the creative draft and selected model carry into the sign-in flow. Generation uses your available credits or eligible plan allowance.
A poster, a product shot, your own photo: one sentence each. Free welcome credits first, 10 credits a picture after that.
Create with Nano Banana 2
