I've been playing with DALL-E 3, and honestly, it's wild how good this has gotten. I asked for a picture of San Francisco with a cat terrorizing the city, and here's what it generated:

The detail is impressive - it got the San Francisco skyline right and made the cat look playfully destructive instead of scary. Here's another variation:

This one even got the trams and the bay area vibe. The quality jump from DALL-E 2 to 3 is huge. OpenAI's really pushing the boundaries here. I'm curious to see how this affects creative industries - not replacing artists, but definitely changing workflows.