Chat Instead of Photoshop: Google's Gemini 2.0 Flash Learns to Edit Images Through Dialogue

Google has taken an unexpected step in the generative AI race by introducing an experimental multimodal version of Gemini 2.0 Flash in AI Studio. The model can not only generate images from scratch but also modify details in existing photos through a simple chat interface — without needing to recreate the entire picture. This marks the first time a major player has introduced such functionality, beating out OpenAI and Elon Musk’s xAI.

Users can give Gemini commands in a conversational format, such as "replace the background with a mountain landscape" or "add a caption in Russian." The AI retains the conversation’s context, allowing for a series of progressive edits. For example, users can change a person's outfit, then "move" them to a different location, and finally adjust the lighting. All changes are applied directly to the original image, saving time and resources.

Every generated image is automatically marked with a SynthID watermark — Google's technology for combating deepfakes. This is particularly important for designers and marketers, who can now legally use the content in commercial projects. Less obvious but equally impressive features include texture cloning, colorization of old photos, and even "filling in" missing elements in images using text-based suggestions.

While Gemini 2.0 Flash understands commands in Russian, the service is not officially available in Russia. Developers and companies can test the model for free through AI Studio or API, but some features, such as face swapping or complex restoration, remain unstable. Google emphasizes that this is an early version, and the final release will be optimized for real-time tasks.

Experts note that Google has, for the first time, combined the creative flexibility of Midjourney, the precision of DALL-E, and the interactivity of ChatGPT into a single model. If the experiment proves successful, it could significantly simplify work in design, education, and even journalism — enabling instant data visualization or creating article illustrations without human involvement. For now, Gemini 2.0 Flash remains an intriguing tool that is already redefining the possibilities of generative AI.