Real or AI? 5 Clues to Spot Images Created with "Nano Banana"
Descripción de la publicación.
1/14/20262 min read


We live in strange times. Two years ago, it was easy to tell if an image was Artificial Intelligence: people had 7 fingers, eyes looked in different directions, and text looked like alien hieroglyphics.
But today, with technologies like "Nano Banana" (Google Gemini) or Flux, the line is blurring. These new AIs can spell correctly, render realistic skin, and understand complex lighting.
Are you being fooled by a fake photo? Here are 5 details you must check to uncover the truth (before you share the news or buy the product).
1. The "Jewelry Test" and Accessories
AIs are incredible at creating the "big picture," but they get lazy with small details.
What to look for: Check earrings, eyeglass frames, or shirt buttons.
The Glitch: Often, an AI-generated person will have a different earring in each ear, the glasses will melt into the skin, or a zipper will disappear halfway up a jacket. Asymmetry in small accessories is the AI's Achilles' heel.
2. The "Dolphin Skin" Effect (Texture)
Although High Definition (4K) is common, AI tends to "idealize" surfaces too much.
What to look for: Zoom in on the face or skin.
The Glitch: Real skin has pores, tiny wrinkles, and imperfections. "Nano Banana" skin sometimes looks too smooth, waxy, or shiny—like plastic or the wet skin of a dolphin. If the person in the photo looks too perfect to be real, they probably aren't.
3. Background Text (The New Detector)
Old models couldn't write. "Nano Banana" can write signs if you ask it to (e.g., a sign saying "SALE"). But it still fails at text that is not the main focus.
What to look for: Don't look at the main sign the character is holding. Look at the shop signs in the background, labels on bottles on a table, or book titles on a shelf.
The Glitch: Often, secondary text remains an illegible scribble or contains letters that don't exist in any alphabet.
4. Light and Shadow Logic
AI understands how to light an object, but it sometimes forgets the laws of physics.
What to look for: Check the shadows on the ground and the reflections in the eyes.
The Glitch:
If the sun is coming from the right, why is the shadow falling to the left?
If you zoom in on the eyes (the pupil), the white reflection (catchlight) should be the same in both eyes (because they are looking at the same light source). In AI, sometimes one eye reflects a window and the other reflects a lamp.
5. Objects "Melting" Together
AI struggles to understand where one object ends and another begins.
What to look for: Hands holding cups, hair touching shoulders, or feet touching the ground.
The Glitch: Sometimes fingers fuse with the handle of a mug, or hair turns into part of a shirt pattern. If you see an object mysteriously disappearing inside another, it is an AI hallucination.
Bonus: Verification Tools
If your eyes fail you, use technology against technology.
Google "About this Image": Google is rolling out features (and digital watermarks called SynthID) to label its own AI images.
Reverse Search: Upload the photo to Google Images. If the photo is of a real historical event, it will appear in many news sources. If it is AI, you probably won't find an original source, or it will lead you to Reddit/Midjourney forums.
Conclusion
There is nothing wrong with enjoying art created with AI like "Nano Banana." The problem arises when it is used to deceive (fake news, product scams, or catfishing). Training your eye to spot these small errors will make you a smarter user who is harder to manipulate.
Contact
Reach out for support or collaboration.
© 2025. All rights reserved.