Thought I’d revisit this with a couple of new AI models. The first is Kadinsky 2.1, available here.
The second is Playground v1, available here.
Strangely, from my limited tries Playground seems to grasp the idea at least a little in 1024x1024 renders but not in 512x512 ones.
1024x1024
512x512
And here is Stable Diffusion 2.1. It does a better job than SD 1.5 up-thread.
Anyone want to burn a few credits testing it with SDXL?
“A garbage truck shaped like an elephant” produced several images of elephants standing near garbage trucks in SDXL.
OpenAI is trying to get ChatGTP 4 to explain ChatGPT 2.
Oh, now that’s cool. I’ll be watching those developments.
Hippopotamus tanks.