/showme v2.0

9

We’ve deployed an upgrade to the /showme command that can be used to create images from text prompts using artificial intelligence.

It works mostly like the previous /showme command. Simply type something like /showme a photograph of an astronaut riding a horse and @mediocrebot will reply in a few seconds with an image.

I’ve changed the AI from an old version of Stable Diffusion (v1.4) to DALL·E. There appears to be an overall quality improvement. The images are higher resolution. In my limited testing the version of DALL·E we’re using doesn’t struggle as much with generating human limbs.

However, DALL·E isn’t great at accurate close-ups of faces (especially of famous people). We also lose the ability to specify the number of steps and the initial seed (not sure anyone was really using those advanced features anyway).

You still need to be a member to use the /showme command. Members are limited to 100 artificial intelligence credits per day. Basically, one /showme command is one credit.

There’s a widget on your account page that shows you how many credits you’ve used for the current day (this only shows up if you use at least one credit for the current day).