AI Algorithm Enhances WC1 Characters (December 28, 2022)

ChrisReid

Super Soaker Collector / Administrator
We've stayed pretty far away from the recent AI art craze, but Patrick Buechner found a phenomenal use for it this week. He's run several of the original Wing Commander characters through the Stable Diffusion package with stellar results. These are an interpretation, of course, and you may or may not like the specific output character, but it's very cool to see a new realistic take on these iconic pilots (yes, Shotglass used to be a pilot "till the fleabags shot me up so bad I couldn’t fly"). Even the backgrounds and choice of collar detail on their clothing is pretty interesting!





It's raining so I'm playing with Wing Commander 1 characters in #stablediffusion #AIart

Patrick has a long resume as an Electronic Arts executive among many other roles in the gaming industry. You can find more on his accomplished background here.

--
Original update published on December 28, 2022
 
Yes, it remains fascinating just what these tools can do, and what they can't. For all the additional details, the AI faces seem less expressive than their 320×200 originals. Although Shotglass, who had the most blank input expression, gains a slight smile.

As you say, the backgrounds are genuinely interesting. I can see how it might turn Shotglass's window frame into a planet and flying saucer. But Maniac's background makes it look like he's the one who's going to be making all the tough moral choices. So I suppose the AI Maniac represents how he imagines himself: as the stoic protagonist whose every choice will determine the fate of entire systems.

Finally, could it just not figure out what to make of that scarf thing they all have? I remember some discussion about which 20th-century US military uniforms the Wing Commander I artists used for reference. There was a time when actual uniforms contained those scarves, right? And there are photographs of people wearing them? But they must amount to a very small fraction of the Stable Diffusion training set.
 
I keep dreaming that in the near future, you could run any old game into an AI emulator and get a completely new experience rendered in realtime.
And then the gnarly nerd in me says, nah, 'twould spoil the experiences you had when playing it the first time.

Nonetheless, I'm intrigued. Does anyone have any idea about the actual text prompt used to create the images? Just "take this and make it prettier" sounds a bit too easy.
 
Back
Top