Google's Gemini AI is really good at watermark removal | The Verge
Google has introduced the latest features of its Gemini 2.0 Flash AI model to developers worldwide. This new AI model now has the capability to remove watermarks from images, among other functionalities. Users have discovered that the Gemini 2.0 Flash AI can efficiently replace watermarks, such as those from Getty Images, with an "edited with AI" mark.
Removing Watermarks with Gemini 2.0 Flash
The Gemini 2.0 Flash AI model, equipped with native image generation, not only generates images from text prompts but also allows for conversational image editing. Users have found that it excels at removing watermarks with precision, as reported by TechCrunch.
While tools like Watermark Remover.io can already remove watermarks from various companies, Google's Gemini AI seems to outperform them, especially in handling complex watermarks like those from Getty Images. After removing a watermark, Gemini 2.0 Flash replaces it with a SynthID mark, essentially substituting the original copyright mark with an "edited with AI" mark.
Concerns and Limitations
Despite its capabilities, there are concerns about the potential misuse of Gemini 2.0 Flash. Users have observed that the AI model can incorporate recognizable images of real people, such as Elon Musk, into photos. However, the full Gemini model restricts this feature.
Currently, developers can access Flash's latest image features through AI Studio. Although the technology is promising, there are questions about safeguards to prevent misuse, such as unauthorized watermark removal. Google has been approached for clarification on the protective measures in place.
Technological advancements like Gemini 2.0 Flash highlight the evolving landscape of AI-driven image editing and the need for responsible usage in the digital space.
