Open AI recently announced that they are giving up on their AI video gen, Sora. Apparently, the project was leaking way too much of their hard earned cash (reports put it at around $1M/day in compute). But recently, they introduced the new image gen model, which they aptly named Images 2.0 (don’t you just love their naming convention), and I decided to put it to the test.
Boulevard du Temple is considered to be the oldest surviving photo featuring a human being. As you can see, the photo is hardly HDR quality, but I wanted to see if ChatGPT could restore it and add some color to it.
As it turns out, this was an easy peasy job for ChatGPT.
Next I wanted to see if it could rebuild architecture from ruins
Very impressive so far. But surely it couldn’t do a cross-section of a human cell
What this new version of ChatGPT Image does is think, plan, and even conduct research before generating an image. But what if….
The CIA is obviously working with OpenAI and forcing them to obscure certain things. But then it hit me: there’s no way it can generate QR codes that actually work
What about Mona Lisa. I wonder how she might have looked in real life
What about the Roman Emperor Augustus
I sometimes wonder what it would be like if were a tiny critter on the streets of New York
But then I was like, “What if I’m just an unsuspecting swimmer, minding my own business, and then all of a sudden a whale jumps out and scares the crap out of me?”
But can it meme?
I then asked it to create a scientifically plausible depiction of the mummification procedure in ancient Egypt.
I have an idea for a social network. Maybe ChatGPT can help me visualize it
But the question remains: is it better than Google’s Nano Banana Pro? To find out, I asked them both to give me a cross-section of a DSLR camera.
I then challenged them to recreate one of the lost ancient wonders of the world, the Lighthouse of Alexandria, and to keep it historically plausible
So which one is better? I honestly don’t know. What do you think? Let me know down in the comments.
Leave a Reply