The all new, all fresh Random Image Thread

Status
Not open for further replies.
8IOS7Dy.png
 
mrk said:
The Jetsons as imagined by Midjourney AI:

<snip>
Click to expand...

I've seen a few of these recently, for various films/shows, and one thing I've noticed is the people (the women in particular) are always very attractive (if you ignore the excessively pointy elbows), and it got me wondering why.

Is the AI "smart" enough to understand that it is generating images for a TV show/film and that Hollywood actors/actresses are generally more attractive and so adjusts accordingly? (In which case how does it "know" what humans would consider attractive?)

Does it take its source material from the internet - e.g. Instagram etc. where people cover themselves in makeup and use filters etc. to make themselves more attractive?

Is it some kind of bias introduced by the programmers, either in the algorithm training, or again the source material provided?

Is it some kind of feedback loop generated by users, where the more attractive images get more engagement, and so it is constantly honing the generation process to improve?

I'll confess to not having a clue how it works, but lets be honest, the characters in the images above - whilst very cool - don't look remotely like the original cartoon, so how/why are they generated in such a way?

jlWQ8mk.png
 
Last edited:
Haggisman said:
I've seen a few of these recently, for various films/shows, and one thing I've noticed is the people (the women in particular) are always very attractive (if you ignore the excessively pointy elbows), and it got me wondering why.

Is the AI "smart" enough to understand that it is generating images for a TV show/film and that Hollywood actors/actresses are generally more attractive and so adjusts accordingly? (In which case how does it "know" what humans would consider attractive?)

Does it take its source material from the internet - e.g. Instagram etc. where people cover themselves in makeup and use filters etc. to make themselves more attractive?

Is it some kind of bias introduced by the programmers, either in the algorithm training, or again the source material provided?

Is it some kind of feedback loop generated by users, where the more attractive images get more engagement, and so it is constantly honing the generation process to improve?

I'll confess to not having a clue how it works, but lets be honest, the characters in the images above - whilst very cool - don't look remotely like the original cartoon, so how/why are they generated in such a way?
Click to expand...
It's all down to the learning model dataset and combined with the actual instructions given to the AI by the person requesting the image to be created. Some key words input into the box likely included the physical appearance of the women, for example.Nobody wants to see a hambeast :p

OChCW0X.jpeg


From: https://www.theguardian.com/science...ho-have-turned-humid-air-into-renewable-power
 
mrk said:
It's all down to the learning model dataset and combined with the actual instructions given to the AI by the person requesting the image to be created. Some key words input into the box likely included the physical appearance of the women, for example.Nobody wants to see a hambeast :p


From: https://www.theguardian.com/science...ho-have-turned-humid-air-into-renewable-power
Click to expand...

Ah yes, of course, the "generate a photorealistic version of the Jetsons and make the chicks well fit" prompt :p

tGLXwJo.png
 
Status
Not open for further replies.
Back
Top Bottom