Try the Menu Visualizer — Free
Point your camera at any restaurant menu. Our AI reads every dish and generates a photo, ingredient breakdown, and nutrition estimate — in seconds. No download required.
Fake food in the window, and then a menu I couldn't read
If you've been to Japan, you know the plastic food models in restaurant windows. They're called shokuhin sampuru. Apollo magazine says they were popularised by Takizo Iwasaki, who set up a factory in 1932, and that the window displays let tourists who can't read kanji see what a restaurant serves. When a place has them, ordering is easy. You point at the plastic tempura and hold up two fingers.
Most of the restaurants I went to didn't have them, and the nicer places almost never do. The menu comes in Japanese, sometimes with no English and no photos. I tried Google Translate and every other tool on my phone to work out what was what. The translation told me what the words meant, but a translated dish name still doesn't show you what's about to land on the table. I ended up guessing more than I wanted to.
Why I built my own
When I got back to Brisbane I decided to build something myself. The idea was simple. I take a photo of the menu, and the app shows me roughly what each dish looks like. It didn't need to be perfect. I just wanted to see the food before it arrived.
I also wanted the story of the dish. When you're in Japan you want to know more about the local food, because that's part of the travel experience. Tempura is a good example of what I mean. The Michelin Guide says Portuguese missionaries brought the method of coating food in flour and frying it to Japan in the 16th century, and Britannica says the dish first appeared in the port city of Nagasaki before it became a specialty of Edo, which is now Tokyo. That's the kind of thing I like knowing before the food arrives.
The last part is a personal one. I keep a very close eye on what I eat, meaning calories, so the app gives you that information too. When you're at a table with no idea what you're ordering, it's hard to make any kind of choice, and I wanted a number to work with before I picked a dish.
How the Menu Visualizer works
The tool is called the Menu Visualizer, and it's free to try in your browser with no download. You take a photo of the menu and it reads the dish names, works out what each one likely contains, and generates a picture. Tap into a dish and you get an ingredient list, an estimated calorie and macro breakdown, and some background on where the dish comes from. It works with most major languages, including Japanese, Chinese, Thai and Italian. English gives the cleanest results.
The scan works better if you follow a few simple rules. Hold the camera flat and straight-on to the page, because angled shots throw off the text reading. Good light matters, since shadows across the menu are the most common reason for a bad read. If the menu is long, scan it a section at a time. In general, stylised fonts and handwriting are harder for any scanner to read, so I'd expect those to be more hit and miss. I go into more detail in my post on the app to scan a menu and show food pictures.
The dish photos are AI-made, so treat them as a guess
The pictures are generated from the dish description and its ingredients. They aren't photos of what that particular restaurant serves, so think of them like the stock photo on a delivery app. They give you an idea, not a promise. A dish can arrive looking quite different, especially in a high-end restaurant where the plating is part of the point.
The calories are estimates too. They're based on how a dish is usually made, not on what that kitchen actually does. Research on AI calorie estimation backs up being careful with these numbers. A systematic review of AI methods found the average error for calories ranged from about 0.1% to 38%, and the errors were smaller for simple single foods. A study covered by Healio in August 2026 found four AI photo calorie apps underestimated energy intake by about a third, although those apps were reading photos of eaten meals, not menus. So I use the number to compare dishes with each other, not as an exact figure. If you have a serious allergy, always ask the restaurant, because an app can't confirm what's in the kitchen.
Other apps that do something similar
I'm not the only one who thought of this. MenuDish says you can snap a menu and see real images of every dish, and its App Store notes mention nutrition info and the origin of each dish. FoodieLens offers real dish photos and translations, and MenuGen generates AI images for each dish from an uploaded menu. The main difference is real photos versus generated ones. Real photos come from other people's plates, so they may not match what your restaurant serves either. I'd try more than one and see which one you like.
What it changed for me
Now a menu I can't read is a lot less stressful. I take a photo, look at the pictures, and read the short background on a dish I've never heard of. When I'm travelling, that's the part I enjoy most. If you like using your phone camera for food, the fridge photo apps I compared and the fruit identifier use the same idea.
Frequently Asked Questions
An AI menu photo generator reads a restaurant menu and creates a professional food photograph for every dish using generative AI. Each image is unique and shows the dish as it would appear on a plate — not a stock photo, but a generated image based on the dish name and description.