Remaking old computer graphics with AI image generation
jalammar.github.io
Remaking old computer graphics with AI image generation
1–10 of 86 posts
Re: Remaking old computer graphics with AI image generation
#2It’s amazing that someday we’ll be able to pass in low fidelity pixel art sprite sheets to an AI and get back high definition hand drawn 2D graphics for use in games.
Re: Remaking old computer graphics with AI image generation
#3It’s amazing that someday we’ll be able to pass in low fidelity pixel art sprite sheets to an AI and get back high definition hand drawn 2D graphics for use in games.
Square Enix: Is it possible to learn this power?
Re: Remaking old computer graphics with AI image generation
#4It’s amazing that someday we’ll be able to pass in low fidelity pixel art sprite sheets to an AI and get back high definition hand drawn 2D graphics for use in games.
you could probably build this w/ current img-gen model tech! totally agree, absolutely crazy.
Re: Remaking old computer graphics with AI image generation
#5It’s amazing that someday we’ll be able to pass in low fidelity pixel art sprite sheets to an AI and get back high definition hand drawn 2D graphics for use in games.
Get out of my head!
I literally just registered spritesheet.ai yesterday.
and I've already had mild success training a model on spritesheet data.
Re: Remaking old computer graphics with AI image generation
#6I have a question. Stable diffusion is based on gradually processing noise into a coherent image, by training a denoiser. Would it be possible to feed low-fidelity image (such as pixel art, or pixelated image) directly into the denoiser step and get a higher-fidelity image that would match the original?
Re: Remaking old computer graphics with AI image generation
#7I have a question. Stable diffusion is based on gradually processing noise into a coherent image, by training a denoiser. Would it be possible to feed low-fidelity image (such as pixel art, or pixelated image) directly into the denoiser step and get a higher-fidelity image that would match the original?
Yes, stable diffusion has an img2img mode that does essentially exactly that.
Re: Remaking old computer graphics with AI image generation
#8It’s amazing that someday we’ll be able to pass in low fidelity pixel art sprite sheets to an AI and get back high definition hand drawn 2D graphics for use in games.
I don't see how animations would work with the current crop of image generation tools. If you feed in 5 frames of a character swinging a sword pixel art you'll get five wildly different renditions of the character, not one character with a smooth tweening of the sword swing.
Re: Remaking old computer graphics with AI image generation
#9It’s amazing that someday we’ll be able to pass in low fidelity pixel art sprite sheets to an AI and get back high definition hand drawn 2D graphics for use in games.
Get out of my head! I literally just registered spritesheet.ai yesterday. and I've already had mild success training a model on spritesheet data.
I'm interested in following your project if you have a blog or anything.
Re: Remaking old computer graphics with AI image generation
#10It’s amazing that someday we’ll be able to pass in low fidelity pixel art sprite sheets to an AI and get back high definition hand drawn 2D graphics for use in games.
Actually, one day we will be able to do this not just with sprite sheets, but whole games.