Halloween 2023: Using AI for Signage
This is silly, but perhaps appropriate. I was trying my hand at creating some “signs” for our Halloween project… and failing. I know what I like, but I’m still just a brute with Photoshop and not really able to visualize the things in my head.
So I cheated, and thought to use DALL-E to generate my images for me.
Our original idea was to print out some kind of banner to put over the porch. But given how expensive that is (and how bad my designs were)… I opted to just play around with some smaller images. And the designs I generated printed out just fine onto an 8.5×11 piece of paper.

First few attempts. I was trying to force DALL-E to use the Creepster font, but that seemed to have mixed results.
After seeing too many hyper realistic images, I asked for some minimal vector images (and got the result on the left). Note that the hand has 4 fingers.

After getting a pretty gruesome example (left), I asked DALL-E to make something a bit more family friendly. And it produced the image on the right. I’m honestly not sure which image is more terrifying.

A little better, and a little less disturbing.

Probably the best of the bunch.
Fun fact: while I had DALL-E create the initial images, I ended up pulling most of them into Photoshop. And did a bit of resizing and cropping. And then also ended up using Photoshop’s generative fill to finish things off.
It feels both silly and appropriate to use AI to fix up an image created by AI. Reminded me of my Infinite Adversaries project, and how I’m using ChatGPT to generate a prompt to send to DALL-E for image creation. Why be lazy, when you can be twice as lazy with AI?
Related:
Halloween 2023: Assembly
halloween 2023: Painting
Halloween 2023: Work Continues
Halloween 2023: Work Begins
Adobe + AI: Photoshop’s Generative Fill
This Post Has 0 Comments