Has anyone tried generating images? I bet a custom system trained for just this purpose would work really well. These are generated images.

Those are impressive, what was the tool and process to generate these?

Those were done with Stable Diffusion 2.1, thay're the result of about 20 different prompts, many of which were nightmare fuel or just plain crazy. Things are going to change...fast.

There's a UMD group for AI WAM, here:

https://umd.net/groups/group/ai-wam