Google’s Gemini is getting new features for editing images with AI. The latest image editing model is the work of the DeepMind artificial intelligence team, and these updates will be available starting today in the Gemini app. Google noted that all images either generated or edited in Gemini will be visibly and digitally watermarked to designate them as created with artificial intelligence.
The first aspect of the new editing model offers consistency across different versions of an image. It’s particularly focused on keeping human characters looking the same as more changes are made the the surrounding visuals, with the idea that people could upload a picture of someone and still look true to reality as they put themselves in different settings or outfits with AI.
The second part of the model involves advanced image editing options. These features include the ability to combine two separate pre-existing images into a new scene, using a visual trait from an existing image as a new prompt or new design element, and multi-stage editing where individual components of an image can be changed in a sequence without losing the accumulated updates.
Image generation of humans was unavailable for a few months last year after Gemini created inaccurately diverse results for prompts based on particular places and times in history. Users regained the ability to AI-generate images of people with the Imagen 3 model.