The definition of photography is rapidly changing as more companies introduce AI tools that redefine how images are created at every step of the photography process, from composition to editing. Does reviving an old photograph by sharpening details and adding colour through AI constitute photography? Who knows, but it’s not stopping anybody from using artificial intelligence. There are 18-year-olds offering “AI courses” for a cool lakh to people who actually witnessed the birth of the Internet but still can’t figure out how to change their WhatsApp display picture.

Take a look at the “Nano Banana” trend that has taken over the Internet. It’s just bananas — literally — what you can achieve. Netizens are posting scaled versions of themselves on transparent acrylic bases, placed lovingly next to their computers, as if they were limited-edition action figures. This is just a small use of Google’s new AI model, Gemini Flash 2.5 Image, previously known as “Nano Banana”. And the best part? You don’t need to hand over thousands of rupees to learn these tricks. That’s the beauty of AI — free chaos for all.

Numbers don’t lie


The new Google DeepMind image generation and editing model can be accessed in the Gemini app. It can turn your pet into a tiny figurine or even a video game character. Google says you can have a tea party with your younger self.

The beauty of the model is that the person or pet remains consistent across edits; you can change a background without suddenly giving grandma three arms. It’s easier to alter parts of an image while keeping the rest the same, and you can apply the style from one picture’s object to another.

Prompt: Recreate this cat as a 16-bit video game character, and place the character in a level of a 2D 16-bit platform video game.

Think of the large stash of photographs lying in your cupboard — grandparents, parents, or your awkward bowl-cut childhood phase. Some are out of focus. Fixing these only requires a Google account and, apparently, zero shame.

Josh Woodward, Google’s vice-president of Google Labs and Gemini, revealed that in just a few weeks, people have created over 200 million AI images, and more than 10 million people have signed up for Gemini. That’s the game: Hook you with free fun, then gently push you towards a paid tier.

Google has also gone all-in on generative media this year. Its AI video generator, Veo 3, has sparked debates with its smooth scenes and synchronised audio. Meanwhile, creators have already churned out over 100 million AI videos with Google’s AI filmmaker tool, Flow.

Prompts you are looking for

This “Nano Banana” craze is getting out of hand. Too many people are using it, and your social feed looks like a toy store catalogue.

Gemini often delivers edited images in under 30 seconds, while OpenAI’s ChatGPT 5 sometimes takes more than three times as long.

The trending prompt to shrink your friends into collectibles is: “Create a 1/7 scale commercialised figure of the character in the photo. The style should be realistic, with clearly defined features, and placed in a real-world environment. The figure should be positioned on an office desk [feel free to improvise], standing on a round transparent acrylic base with no text. On the computer screen, display the Adobe Illustrator modelling process of this figure [once again, feel free to improvise]. Next to the screen, place a Ken-style toy packaging box [can be improvised] printed with the original photo.”

Next, grab a photo, copy it, paste the prompt, hit enter, and voila — within 30 seconds you’ll have something bizarre to terrify your friends on Instagram. Even the free version of Gemini plays along.

Another popular prompt: “Replace the man with a buffalo and tweak the lighting to make the feel cooler.” It works like a charm — provided you’ve always wanted your wedding album to look like a National Geographic special.

Where the new model stumbles is with faces. They sometimes look a little too smooth, like everyone’s been Photoshopped by a skincare brand. Maybe it’s deliberate — to remind you this is AI and not your actual aunt.

Old family photos? Throw them into Gemini, ask it to colourise and sharpen them, and you’ll either get a faithful restoration or something that looks like your grandfather moonlighting as a Marvel villain.

Feeling hungry? There’s a prompt for that too: “Turn these ingredients into a refined delicious-looking dessert, inspired by these ingredients. Plate it as if it were a dish at a 5-star avant-garde restaurant.” Because why cook when you can hallucinate a Michelin star?

How do you know it’s AI?


It’s becoming harder to tell AI-generated photos from real ones, especially when they flash past on your feed in under two seconds. Google’s model sometimes leaves random details untouched, which might be your clue. There’s also a tiny indicator in the bottom-right corner to show AI’s contribution, but of course, that can be cropped faster than you can say “Photoshop”.

Google does embed “SynthID” watermark data in AI-edited images, a kind of invisible tattletale. But even they admit it’s not foolproof. Last year, Google said: “SynthID isn’t a silver bullet for identifying AI-generated content, but is an important building block for developing more reliable AI identification tools and can help millions of people make informed decisions about how they interact with AI-generated content.”

Prompt: Colourise the attached photo, including the background. Correct focus. Get rid of the patches from the photo.

As Samsung’s EVP Patrick Chomet put it to TechRadar in January 2024: “Actually, there is no such thing as a real picture. As soon as you have sensors to capture something, you reproduce it, and it doesn’t mean anything. There is no real picture, full stop.”

Comforting, right? Meanwhile, Google Pixel Camera leans into the confusion. Isaac Reynolds, group product manager, told Wired in August 2024: “It’s about what you’re remembering… A perfectly accurate photo can feel fake. Edits help you create a version of the moment that feels authentic to your memory, even if it isn’t authentic to the millisecond.”

Translation: Reality is overrated — and memory is now editable.