How much does it cost?

Each image costs 10 to 13 credits depending on the model, the same for text to image and image to image. Credit packs are one-time purchases from $2.99 for 100 credits, so an image costs roughly $0.15 to $0.39. Credits are valid for 365 days and there is no subscription.

Do the images have a watermark?

No visible watermark or logo is added, and DeGemini never puts the Gemini sparkle on your images. Like most AI providers, some models embed invisible provenance data (Google's SynthID, for example), which doesn't change how the image looks.

How long does a generation take?

Usually 10 to 60 seconds. Keep the page open; the image appears as soon as it's ready. If a generation fails, the credits are refunded automatically.

What happens to the photos I upload?

Reference images are sent to the AI provider only to run your edit and are deleted automatically within 24 hours. Only upload images you have the right to use, and never photos of other people without their consent.

Which files and sizes are supported?

PNG, JPG and WebP, up to four reference images per edit. Large images are resized to 2048 px in your browser before upload. Outputs come in 1:1, 3:2, 2:3, 16:9 or 9:16, plus Auto on Nano Banana 2 and GPT Image 2.

Do I need an account?

Yes, a free account is needed so your credits and images stay yours. The Gemini watermark remover on the homepage stays free and needs no account.

4 models · up to 4 reference images

Image to Image AI

Upload a photo, describe what should change, and get a new image in seconds: restyle it, swap the background, or combine up to four references. No sparkle watermark, ever.

Image generation costs credits per image (the price varies by model), so we need to know who's spending them. Sign in to continue — the Gemini watermark remover on the homepage stays free.

What is image to image AI?

Image to image AI creates a new picture from one you already have. Instead of starting from a blank canvas, the model reads your reference photo, keeps what you want kept (the subject, the pose, the layout) and changes what your prompt asks for: the style, the background, the lighting, the colours or the details.

DeGemini gives you four of today's strongest editing models in one place: Google's Nano Banana 2, OpenAI's GPT Image 2, Black Forest Labs' FLUX 2 Pro and ByteDance's Seedream 5 Pro. You can add up to four reference images per edit, for example a product shot plus a background, or a character from several angles, and every result is delivered without a visible watermark or logo.

How to use image to image AI

1
Upload your reference images
Add up to four PNG, JPG or WebP images: drop them in, paste them, or click to browse. Big photos are resized to 2048 px in your browser before upload.
2
Describe the change
Write what should be different, for example "turn it into a watercolor painting" or "put the product on a marble counter in morning light". Name what must stay the same, too.
3
Pick a model and aspect ratio
Choose Nano Banana 2, GPT Image 2, FLUX 2 Pro or Seedream 5 Pro. Auto keeps the shape of your first image on Nano Banana 2 and GPT Image 2.
4
Generate, refine, download
Most edits finish in 10 to 60 seconds. Not quite right? Click "Edit this image" to feed the result back in, then download the file in its original format.

Which model should you pick?

All four work for both text to image and image to image. Credits are charged per image and refunded automatically if a generation fails.

Model	Best for	Credits per image
Nano Banana 2	All-round edits that keep faces and products consistent; fast	13
GPT Image 2	Product shots, clean layouts and legible text in the image	10
FLUX 2 Pro	Polished, commercial-style graphics and illustrations	10
Seedream 5 Pro	Photorealism and poster-grade typography, including Chinese	12

What people use it for

Restyle a photo as a watercolor, anime frame, oil painting or 3D render.
Replace or extend the background behind a person or product.
Turn a plain product photo into a lifestyle or studio shot.
Keep a character consistent across scenes using several references.
Change colours, outfits or text, or remove unwanted objects.
Turn a rough sketch or wireframe into a finished image.

Prompt tips for better edits

Say what must not change: "keep the face, pose and framing exactly the same".
Ask for one main change per run, then build on the result with "Edit this image".
With several references, say which image provides what: "the person from image 1, the jacket from image 2".
Describe light and lens ("soft window light, 50 mm, shallow depth of field") for realistic results.