Gemini App Image Editing Is More Widely Rolling Out

Following last week’s announcement, native image editing in the Gemini app is now seeing wider availability.

Previously, if you had Gemini generate an image and then asked for a change, an entirely new picture was created:

Old (Different dog, background, clouds, etc.)

Native image editing keeps the base image while making specific changes via natural language prompts. This works for both generated images and ones you directly upload to Gemini. With a text prompt, you can change backgrounds and styles, replace objects, and add elements, including text.

In the example below, the second image is identical (including the dog, trees in the background, sky, etc.) to the first except for the blue grass.

New

You’ll also notice how generated images now feature a visual “ai” watermark in the bottom-right corner. This is something Google is testing, and it joins the invisible SynthID digital watermark.

Google also says “you can upload a personal photo and prompt Gemini to generate an image of what you’d look like with different hair colors.”

This multi-step editing, which preserves context throughout the conversation, also makes possible ”prompts with text and images integrated.” Examples of that include step-by-step instructions with images, or a “first draft of a bedtime story about dragons and provide images to go along with the story.”

It’s powered by Gemini 2.0 Flash, but you can invoke this tool/capability in any model. As of today, we’re seeing wider Gemini app availability of image editing in the US across several free and Advanced accounts.

For developers, Google announced today that Gemini 2.0 Flash Image Generation and editing is now in preview for developers. gemini-2.0-flash-preview-image-generation is available via the Gemini API in Google AI Studio and Vertex AI today. Compared to the March test, Google touts improvements like: