When traveling in Europe, ChatGPT has become my most reliable personal tour guide. Walk into a museum or a church, snap a photo and send it over, and it can walk me through the history and stories behind what I’m looking at. Of all the large models out there, ChatGPT might not be the strongest at every single thing, but taken as a whole, it’s hard to find anything that really replaces it.
ChatGPT has a knack for landing on the most fun ways to use its own features. Today’s “turn a photo into a cartoon” feature is a great example: hand it a photo, say something as simple as “turn this into cartoon style,” and you get back a genuinely impressive result — one you can keep revising from there.
Idea 1: Turning a photo into a cartoon#
This is a photo from my mom’s birthday a few days ago. I asked ChatGPT to turn it into a cartoon. What came back was a detailed, warmly colored cartoon image that still felt true to the original.
Idea 2: Editing details within a generated image#
In this generated image, the clothes the kid was wearing got misread as a Mao suit.
I told ChatGPT the boy in the middle was actually wearing a suit, and it immediately regenerated a more accurate version — the clothing details came out looking natural.
Idea 3: Adding elements into a photo#
This is a photo of us on a beach in Tanzania. I asked ChatGPT to add a US warship in the distance.
It generated a scene with a warship visible far off on the horizon, and the result looked convincingly real.
Here’s a leopard I managed to photograph on the African savanna — a rare shot. I asked ChatGPT to put the leopard face to face with a dinosaur.
The leopard’s pose ended up looking a little too relaxed for the situation, but ChatGPT did faithfully complete the “add a dinosaur” part of the request.
Here’s my son getting a close look at a giraffe.
This time it generated a dinosaur hiding in the grass.
Idea 4: Generating a new setting from a portrait#
I also sent ChatGPT a photo of myself cycling around the neighborhood and asked it to restyle me in a military look. Not only did the outfit change, but the whole bearing of the person in the photo came out noticeably more rugged.
Older image-generation models felt more like VR — whatever they generated was its own separate world. ChatGPT-4o’s new image feature feels more like AR: it takes a real photo and drops you into a world you created yourself.
Idea 5: Generating a photo of two people together#
Give it individual portrait photos of two people, and with the right prompt it can generate an image of them together.
Right now ChatGPT can’t produce a genuine joint photo — it can only convert the result into a cartoon style.
A few more photos#
That winter, building a snowman and having a snowball fight downstairs. That year, hiking together. This year’s family portrait.
