No, I mean the accusatory women and the cat kvetching about the green food.
Apologies for misunderstanding. It’s referring to this meme, which has been very popular over the past few years:
Oooooh! I should have caught that!
Someone should do “distracted boyfriend” meme. Not sure how to make that prompt.
I have this saved as a Photoshop template if anyone else wants lego meme text added.
Best “distracted boyfriend” I’ve made. I think it’s not really possible to recreate the original.
I actually like your one-image version better. Hews closer to the original - the women are accusatory, angry, the cat calm and defiant.
I entered “Orson bean with a stoat and a bodkin”
and got the following message…
Unsafe image content detected
Your image generations are not displayed because we detected unsafe content in the images based on our content policy. Please try creating again with another prompt.
OMG What have we done ?!?!?!?
It kind of randomly gets upset. Just rephrase somehow. I’ve had prompts that couldn’t possibly be offensive get rejected. It even changes from day to day.
The fascinating (to me) thing about this is that nobody sat down and tried to write a program for creating Lego images. They just threw billions of images at a neural network and said “learn these patterns” and enough of them were Lego images that it figured out how to make new (imperfect but pretty good) Lego images on its own.
Bing filters out a list of celebrity names. Orson Bean must be big enough to make the list.
Good catch, I missed the proper name thing. It really just doesn’t like surnames at all, doesn’t matter if it’s a “real” name.
SDXL isn’t as good as Dall E 3 with understanding prompts or often with image quality, but it did a good enough job here, and doesn’t mind doing famous people (It is up to the viewer to determine what the pantless Orson Bean is doing to the stoat.)
You can use SDXL here (among other places)
“LEGO squeegee creating memes”
Oh well, let’s leave out Orson…
Here is “New year’s eve at the Stoat and Bodkin in Legoland”
I’m not sure which one is Orson Bean. Or Tami, for that matter.
Ha!
Darren’s been doing yeoman work too!
Well, hell. All I had to do was put in ‘Lego “distracted boyfriend” meme’ and this came right out.
One of the interesting quirks of Bing/DE3 is just how very different two runs of the same prompt can be. I just did lego “distracted boyfriend” meme and got this set.
This happens a lot, outputs of four images with specific details related to each other but not related to other runs of the same prompt. I’ve mentioned this before in the main AI images thread in CS, and speculated on that coming from the ChatGPT they use injecting extra descriptive text to prompts “behind the scene”.
Yeah, I ran that prompt a few more times, hoping to get an accusatory look from GF on the right, and got nothing at all like the image I posted. People walking around covered in hearts, random street scenes, a guy angrily glaring at a lesbian couple. It’s just random as hell. Yet somehow I got the Meme image on my first try.
‘lego “woman yelling at cat” meme’ has so far yielded a bunch of nonsense. I mean, it’s meme-able to a point but I was hoping for something like your image.
Yeah, I tried asking for the cat meme after today’s exchange, too. Nothing accurate, but some pretty funny.
But what I came to post was that I tried “a funny meme recreated with lego”, and look at what happened with my first try.