Google is enhancing its Gemini app by introducing an artificial intelligence (AI)-driven native image editing feature. This development, announced on Wednesday, enables users to modify images directly within the AI chatbot, streamlining the editing process. The feature was initially revealed in March as part of the Google AI Studio platform and relies on the advanced Gemini 2.0 Flash AI model, empowering users to interactively request changes, add or remove objects, and enhance backgrounds, among other edits.
Gemini App Now Supports Native Image Editing
While Gemini entered the image generation arena later than competitors like ChatGPT, Copilot, and Midjourney, the feature was initially rolled out in early 2024. However, it faced challenges when Google had to halt its deployment due to users encountering inaccurate and inappropriate image outputs. The image generation capability was restored in December 2024 with the launch of Gemini 2.0 Flash.
According to a blog post from Google, the native image editing functionality will be gradually rolled out to users worldwide, supporting over 45 languages. The rollout will occur in phases, meaning it may take several days for all users to gain access to the feature.
This Gemini 2.0 Flash-enabled tool allows users to either edit AI-generated images or upload their own photos for modifications. For example, someone could provide a photo of a table and request that the AI add flowers to it.
Users can also challenge Gemini with more intricate editing prompts, such as altering the background, substituting objects, or making detailed adjustments to the subjects depicted. An illustrative scenario includes uploading a personal photo and asking the AI to visualize how the user would appear with a different hair color.
However, there are concerns regarding the potential increase in deepfake technology. In response, Google states that its AI models are designed to reject any requests deemed offensive or potentially harmful. Furthermore, images that are created or modified using the native image editing feature will include an invisible SynthID digital watermark, with the company also exploring the addition of a visible watermark on all AI-generated images.
