Google Launches Nano Banana Pro for Realistic AI Ads

The user provided the instruction, "Arrange the living room using these furniture items," and shared individual images of a sofa, ...

The user provided the instruction, "Arrange the living room using these furniture items," and shared individual images of a sofa, bookcase, coffee table, and rug to Google's artificial intelligence (AI) model 'Gemini.' The 'Image Generation' and 'Thinking Mode' functions were activated.

Five seconds later, the AI named 'Gemini' generated an image of a living room that included all four pieces of furniture. The carpet's design and hue were identical, and the items on the bookshelf from the original picture were also exactly duplicated. In addition to the user-supplied furniture, the AI added slippers, a plant, and some snacks on the side table, resulting in a realistic photograph similar to those seen on online furniture retailers.

Google has recently introduced 'Nano Banana Pro,' an image creation and editing tool powered by 'Gemini 3 Pro,' considered one of the most sophisticated AI models available today. This upgraded version expands upon the success of 'Nano Banana,' which became well-known for its high-quality image generation features similar to those found in Photoshop.

Utilizing the most advanced reasoning features, the tool enables users to create and visualize concepts, generating images that are virtually identical to actual photographs and of sufficient quality for use in major advertisements.

The variety of images that can be produced or modified with 'Nano Banana Pro' is extensive. It has the capability to generate highly realistic images that exceed the quality of high-definition photographs. The software generates images in resolutions ranging from 2K to 4K and allows for custom resolutions based on user requests. Moreover, users have the ability to adjust ambient lighting, camera perspectives, focus, and color correction, as well as apply advanced camera effects such as focus transition. This removes the necessity of using 'Photoshop' or taking manual photographs to create or design visually natural advertising images.

For instance, when requested to create an image of a "glass bottle holding a red pill, lit by sunlight," the AI accurately depicted the shadows on the ground and the reflected light.

The tool's capacity to create images containing text has also enhanced. In the past, text in images frequently looked distorted or abstract, but 'Nano Banana Pro' correctly displays advertisements, menus, and signboards with text. Its processing of the Korean language, which was previously less effective than English, has also seen improvements. It now translates English-based products, advertisements, and menus into Korean without mistakes.

The tool ensures uniformity across as many as 14 images and up to five distinct facial features. For example, if 14 images of plants are submitted with the request, "Combine these plants to create a garden," 'Nano Banana Pro' can produce the scene. Likewise, if pictures of five different people are provided along with the instruction, "Generate an image where they take a selfie together at the beach," the tool can create matching faces for all five individuals.

When presented with a scene from the well-known variety program 'Yoo Quiz' and instructed to create an image of "U.S. President Donald Trump interacting and speaking with Yoo Jae-suk and Cho Sae-ho," the tool generated a realistic photograph that looked authentic.

Furthermore, 'Nano Banana Pro' produces images using real-world data. It is now capable of generating data-based visuals like biological or scientific illustrations, geographic details, and charts. For instance, users can enter a detailed scientific article and ask, "Visualize this paper."

Following the launch of 'Nano Banana Pro,' the developer and design communities have been overwhelmed with instances of image creation and modification utilizing the tool.

As AI-generated images grow more advanced, worries emerge regarding their possible abuse in criminal activities, copyright violations, or breaches of portrait rights. Addressing these issues, Google has introduced a simple verification system through a chatbot to determine if an image is AI-generated. A concealed digital watermark known as 'SynthID' is included in media produced using Google's tools. By uploading an image to the Gemini app and asking, "Is this AI-generated?" you can confirm its source.

Nevertheless, even with increased transparency via watermarks, concerns remain that these extremely realistic AI-generated images might still be misused for deceptive purposes.

Google intends to first introduce this feature for images and then extend it to audio and video. Visible watermarks will continue to be present on images produced by free and 'Pro' users, while 'Ultra' subscribers will have the option to create images without visible watermarks.