This brand new image generation model from OpenAI is lightyears ahead of version 2. To give you an idea of what I mean, here’s a comparison using the same prompt.
“Drawing of a playful monkey, wearing a space helmet and suit, riding a majestic horse across a lunar landscape. The monkey has a bagel in one hand, taking a big bite, with moon craters and a distant Earth in the backdrop.”
Dall-E 2
Dall-E 3
Well it’s quite a marked improvement to say the least! While I can appreciate the childlike stylings of v2, I haven’t come up with many practical applications for it.
V3, on the other hand, has some serious potential!