is dall-e the weird one or is it just me
hello everyone today i put really funny prompts into the dall-e image generator ai and got some really awesome results. lets start off with the first one:
what i input: “a person drinking soup at night”
what i got:
![](http://ilsenotisle.com/wp-content/uploads/2023/02/DALL·E-2023-02-15-12.06.51-a-person-drinking-soup-at-night-1024x1024.png)
i started off with a relatively simple prompt and got a couple of images, but this was my favorite one. i think it hit the prompt, but of course, it’s not perfect. first off- that spoon is NOT a spoon. i dont even think you can drink soup using that thing? the person is pretty realistic looking, but the thumb holding the bowl? that’s no human thumb. he is not a human. the eyes don’t help at all.
…and i’m not even sure if i should point out the two orange dots in the back. there’s gotta be some feral creature hiding there.
alright, next prompt.
what i input: “mickey mouse ascending into the skies in a bright glowing light”
what i got:
![](http://ilsenotisle.com/wp-content/uploads/2023/02/DALL·E-2023-02-15-12.09.47-mickey-mouse-ascending-into-the-skies-in-a-bright-glowing-light--1024x1024.png)
what did they do to my guy mickey mouse?!! what kind of evil creature is this?? they hit the “ascending into the skies in a glowing bright light”, but it appears they didn’t really understand what a mickey mouse is. either that, or dall-e isn’t allowed to use mickey mouse in its ai generation because of disney destroying them, but a couple of the other generations had a better looking mickey mouse. none of them hit him exactly, though.
he kind of seems like some kind of five nights at freddy’s animatronic here? why is his hand and foot like that.
next prompt..
what i input: “squids invading the eiffel tower and there’s an anime girl watching them and crying”
what i got:
![](http://ilsenotisle.com/wp-content/uploads/2023/02/image-1024x391.png)
i had to include all four of the photos. i have never been so shocked at an image. dall-e clearly struggled to figure out how to put an anime girl in, but i’m really surprised it couldn’t generate better looking squids. and the third image has manta-ray looking creatures. the fourth one i won’t even talk about.
and yet again, dall-e struggles with hands.
next prompt!
what i input: “george washington communicating with the aliens while standing on the moon”
what i got:
![](http://ilsenotisle.com/wp-content/uploads/2023/02/DALL·E-2023-02-15-12.15.35-george-washington-communicating-with-the-aliens-while-standing-on-the-moon-1024x1024.png)
unfortunately, he’s not communicating with any aliens. in fact, there are no aliens at all! it appears he is on the moon though, and i like the stars kind of representing america but also the stars in the sky. very interesting one here. but again, dall-e can’t seem to get hands right.
our last one
what i input: “draculaura from monster high in the style of starry night”
what i got:
![](http://ilsenotisle.com/wp-content/uploads/2023/02/DALL·E-2023-02-15-12.17.38-draculaura-from-monster-high-in-the-style-of-starry-night-1024x1024.png)
i asked for draculaura and for some reason got an entirely new vampire-looking character. she’s pretty cool, and i like the space-feeling i’m getting from it, but it’s not really what i asked for? it slightly reminds me of starry night, but not fully.
and also, again- the hands.
dall-e really doesn’t like hands. that’s really the big thing i noticed in all of them. but as someone who spends far too much time online, i’m well aware of the issues image ai generation can bring.
i’ve seen someone posting ai generated ‘drawings’ and claiming them as their own, saying they made it themselves. people’s art gets stolen and altered using ai image generation, and while you can tell when it’s ai generated at times (cough cough, the hands tend to ruin it), the more we feed ai these prompts and images, the better it will become over time.
it can be great for thinking up concepts or drawing inspiration for content, but it could easily be taken by anyone average person and used to trick others into believing it’s theres. the last image i generated could be used to create an original character concept! but i could also take that image, post it online claiming its mine, and sell my “art” for money despite the fact that i put no work into it.
and yeah, that’s pretty much it. image generation could easily be used for good, but just as easily for bad. it really depends on the person using it.