Gemini will soon allow you to fine-tune your generated images (APK teardown)

Image generation is one of the core features of any AI chatbot, and there’s no exception for Google Gemini on that front. As with other chatbots, Gemini allows you to create images from text prompts and then use those images for whatever you like. However, one thing Gemini lacks is the ability to tweak an image after it has been generated — but that could change very soon.

In the latest Google beta app for Android (v15.29.34.29 beta), we see new tools related to fine-tuning AI-generated imagery. This could be incredibly useful for situations in which you like the image that’s been created but don’t like a particular detail about it. Today, Gemini has no way to fix an image — you just need to create a new one through a new prompt, which means you could spend a ton of time going back and forth with Gemini trying to get the right outcome.

This new fine-tuning feature, though, will make this a lot easier through two different methods. First, Gemini will now be able to keep the core of an image but change certain things through natural language prompts. In the example below (middle image), you can see how Gemini will understand that you don’t want a new image, you just want to change the one that’s already there:

The second method for doing this is to use your finger (or stylus on supported phones) to circle an aspect of the image you want to change (screenshot above, right). You can then tell Gemini what you want to edit without needing to describe the part of the image to which you’re referring — Gemini will understand that you’re fine-tuning the circled part.

Unfortunately, although we were able to see the evidence for these features, we weren’t able to get them to work properly. It’s likely Google will need to do some server-side switching before they will become active. Still, it’s nice to see that there could be a lot more control over image generation coming soon to Gemini.