The menu changes every day. The photo rarely does. For years, kitchens accepted they couldn't have good photos of their food. Guests ordered at the counter, so they didn't need them. Self-service kiosks and apps changed that.
If you look at the menu in most workplace restaurants before lunch, whether it's in the app, on the screens or at the kiosk, you'll usually see one of three things. A generic picture featuring a selection of Italian dishes. A stock image of someone else's lasagne. Or an empty placeholder where a photo should be.
It's not because chefs don't care, quite the opposite actually. Most chefs I meet care a lot about how their food looks. They plate it carefully, and then the photo on the screen shows something else. The problem is the maths and economics.
And it costs something. On Just Eat, menu items with a photo are four times more likely to be added to the basket. DoorDash says restaurants have seen 44% more monthly sales with item photos.
The menu changes every day
A chain like McDonald's or Five Guys runs largely the same menu in every restaurant, so one photo shoot covers every location for a season. In contract catering that doesn't work. A contract caterer cooks for each client, not just for guests. One client wants a plant-forward programme, another has a tight budget, another has staff from many food cultures. And the same people eat there every day, so the menu has to keep moving. It often changes every day, sometimes on the day itself. The greens change. The garnish changes.
So a photo is only right for a specific location, on a specific day. And it matters more than it looks. Satisfaction is won or lost in the gap between what guests expected and what they got, and the photo sets that expectation. If the screen shows one dish and the counter serves another, the meal starts with a small disappointment, even when the food is good. Delivery apps have seen this in their numbers: when restaurants on Zomato used AI images that didn't match the food, its CEO said refunds went up and ratings went down. In workplace dining the same guests come back every day, so the gap adds up.
But anything that takes more than a few minutes, or costs a lot of money per dish, doesn't fit. That's why most dishes go out with the wrong photo, or none at all.
An evolution of food photos on menus: what's changed since the 1970s
Every new way of getting food photos onto a menu was built for someone else: chain restaurants, advertisers or people selling things online. All of them assumed the menu stays the same for a while. That's why none of them has fitted contract catering well.
Through all of it, the photo and the plate have rarely matched.
Why none of them fit a daily menu
Photographers: good photos, but not every day
A good food photographer makes a dish look as good as it tastes. But you pay for the day, and a dish might be on the menu once. Nobody books a shoot every morning, and it’s quite expensive.
Stock photos: cheap, but not your food
Cheap and quick, which is why they're everywhere. But guests see one lasagne on the screen and get another at the counter.
Lightboxes: better light, more work
The light is better and more even. But they take up space on the counter, someone has to carry each plate over before service, and on the day nobody charged the camera, there's no photo at all. The photo still has to be uploaded and added to the menu.
AI-generated images: fast, but not the real dish
Fast and polished, but the AI decides what the dish looks like, not the kitchen. Many still look cartoonish, and when the image is wrong, guests are the first to notice.
Worth a thousand words
A menu line can say "roast chicken, lemon, herbs". A photo shows whether it's the dish you want today. Guests decide what to eat before they reach the counter, in the app, on the screens or at the kiosk, and the photo is often the first thing they see.
Delivery data points the same way. On Just Eat, a menu item with a photo gets 16% more orders on average. There's also a simpler reason. Chefs work hard on the food. It should be seen as it is.
Why that's changing: AI should change the light, not the food
The question isn't whether to use AI for food photos. It's what you let it do. Generating a dish from a menu description means the AI decides what the food looks like.
Starting from a photo of the real dish means the kitchen decides, and the AI only works on the photo.
We think the second is the right job for AI in food service. Change the light, the background and the framing. Leave the food alone. If it's not on the plate, it's not in the photo.
That gave us a few rules for what we built:
Start from a photo of the real dish, taken in the kitchen.
Don't add, remove or change any food.
Let the kitchen choose how it looks, and show the original next to the result.
Make it quick enough to do every day, on a phone.
Put the photo straight on the menu, with no separate upload.
How Kanpla Photo Studio works
Kanpla Photo Studio is a tab in the Kanpla Admin app. A chef takes a photo of the dish on their phone. While it's processed, they choose where it goes: today's menu or the product bank. They pick the angle, the background and the container, check the result against the original, and publish. The photo is then live wherever Kanpla shows the menu: the app, the kiosk and signage.
Pick a style once and every site can use it, so the menu looks the same across kitchens.
A photo that looks good and is true
Every option kitchens have had so far asked them to choose between a photo that looks good and a photo that is true. We don't think they should have to choose any more.
If you run a kitchen and want to try it on your own food, you can contact us at Kanpla. And if you have views on AI photos of food, good or bad, I'd like to hear them.
