🎃 spooky inspiration

Things move fast with ai, and at the end of April, stability.ai released DeepFloyd IF.

What this ‘powerful text-to-image cascaded pixel diffusion model’ does is generate images with words that you have actually seen before (it does other things, but I’m focusing on the text-generation).

screenshot from the announcement page, notice the ‘capybara podcast’

What about the spooky-ipsum?

I like to use the old 1.4 model to generate images, which means the text that comes out is often bizarre.

I like it.

I thought some Halloween posters from a combination of the below artists might be interesting: (links go to their wikipedia)

• ray eames
• s. neil fujita
• morton goldsholl (wiki goes to Morton Goldsholl Associates)

The prompt was pretty basic:

a spooky, colorful Halloween poster, by Ray Eams, S. Neil Fujita, and Morton Goldsholl

and I combined it with the ‘universal negative prompt’

ugly, titling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face

this is a great article that contains way more info than just negative prompts

Spooky results

note: i’d avoid trying to utter any of these incantations until further research can be done 😉

i’ve gotten some inspiration and ideas from these (from the color palettes to the bizarre faces & all the strange details in between).

if you’re generating images and haven’t experimented with negative prompts, i’d read this article and give it a go.

i hope you’re doing well wherever you are.

best,
Kelsy

👇 partII

🎃 spooky inspiration, ptII