Nano Banana creates, but can also listen to the planning that Gemini tells you to do
What you need to know
- A tipster found evidence in Google’s code that suggests the editing feature is specific to Gemini’s Nano Banana creation app.
- Reportedly, this feature will be introduced with a pencil icon on photos, giving users the opportunity to “Mark” the changes they want by drawing or typing.
- Rumors from a few months ago alleged the same power to users through the Gemini app with the “resend” option.
There may be new power directed at Google’s popular Nano Banana image-generating software in Gemini.
A report from AssembleDebug with Android Authority says that Google is working on a way for users to quickly edit Nano Banana creations. The tipster suggests that the main steps to get the Nano Banana to produce the image remain the same; however, change once the result is produced. The post says the Gemini app will display a pencil icon, similar to what you might see for editing purposes in other apps.
On tapping, users can be taken to the “Mark” screen. From there, the tipster shows Google’s progress in improving the “Sketch” and “Text” options, next to your pencil colors at first. It appears that users can circle or highlight an area that needs a change. After that, the image is sent back to your chat window, as it awaits your text description of what Gemini should change.
The tipster’s example reads, “change the color of this to green, too.” With that, it looks like Google’s AI will take your written description and what’s circled or highlighted into account, so users won’t have to (literally) repeat themselves. The AI image generator (Nano Banana) will take an image, apply the transformation, and return the result.
A feature it seems compared to meat; However, the findings of AssembleDebug did not decrease when users saw this coming to the application.
Same discovery
What a tipster recently discovered in the latest version of the Google app is similar to rumors from last year. Earlier reports claimed that Google was working on an “annotation” for the Gemini app. This was about what images the AI creates using its generator. Users can “write or draw” items on their photos before downloading them, alongside a possible “repost” option.
That redirect was said to allow users to make changes or additions before adding instructions for the AI to complete. The recent acquisition of AssembleDebug is very close to this, which means that Google has not completely abandoned this idea and is still working on it.
Android Central Take
There is little QoL here with this rumored feature. Right now, if I want to change something, I’ll have to tell the AI and be very specific, hopefully it will understand what I’m talking about. With this leaked method, users can specify in the drawing itself; actually pointing out to the AI that “this place here must be different.” Combining this with your written instructions ensures that the AI gets it right. It’s useful, and will probably help users think more about ideas before bringing them to life.
