I tend to have a lot of silly ideas. Not useful ideas. Not good ideas. Silly ideas. Randomly yesterday I was thinking about the Twelve Days of Christmas song. If you aren't familiar with it, it starts off with a gift for one day, then repeats and adds a second day, and so on and so on. The gifts are:
partridge in a pear tree
two turtle doves
three French hens
four calling birds
five gold rings
six geese a-laying
seven swans a-swimming
eight maids a-milking
nine ladies dancing
ten lords a-leaping
eleven pipers piping
twelve drummers drumming
I thought - what if I took each of these phrases and dropped them into an AI image generator? I did, and the results were... kinda fun. Before I show them, some notes.
- Initially, I did not alter the text in any way, except to sometimes swap out the numerical word ("nine") with a digit (9). Pretty much every service I used had trouble with showing X items of something. I decided to not fight it and just go with it.
- Nearly every service blocked
eight maids a-milking. I'm not really sure why. Eventually, I found that this prompt generally worked "eight young women working with cows on a dairy farm".
- In general, I picked the best image out of the options I got, but obviously, that was my personal opinion.
Ok, let's take a look!
Imagine with Meta AI #
My first set of images comes Meta's Imagine, which I believe launched just a day or so ago. As far as I can tell, it's free to use, but watermarks the images.
Google Search #
The next set comes from Google's generative search in their search results. Honestly, this one was a bit wonky. First, you have to enable AI in search results for your account. Second, you have to, in search, ask for a drawing, so for example, "draw a picture of a cat looking smug", and lastly, it only works in Chrome, which frankly is stupid. That being said, when you used a generic prompt like the above, the results would 'riff' on them and do things like, "an oil painting in cubit style of ..." which was kinda cool. Unfortunately, I couldn't get the 'maids' thing working, even with my alteration, and had to go to "people", not "women", and it refused to do dancing ladies at all, so you'll only see eleven results result.
Adobe Firefly #
Quick disclaimer, I work for Adobe, although not on the Firefly team. You can try it out at firefly.adobe.com if you've got a Creative Cloud membership, and the tech is also used directly within Photoshop as well.
ChatGPT DALL-E 2 #
This is not the current version of DALL-E available for ChatGPT users (see the next section), but was available "for free" at labs.openai.com. Free with limited credits, but I had enough to generate my set.
ChatGPT DALL-E 3 #
For the most recent version of DALL-E, I've got my coworker Peter Nam to thank. I'm too cheap to pay for ChatGPT myself, but he is not. :) You will notice that both this set, and the next set, picked up on the Christmas theme without explicitly stating it, which is really freaking cool.
Image Creator from Bing #
For the final set, I used Image Creator from Bing. This is free, but they will "slow you down" after so many uses. (I didn't hit that while testing.) As with DALL-E 3, Bing picked up on the Christmas theme, and honestly, the first result is just plain stunning.
Merry Christmas! #
I hope you enjoyed this little experiment, I know I did. I'd love to see others give this idea a try as well, and if you, do reach out and let me know!