foone , Englisch
@foone@digipres.club avatar

Novel compression method: we take all the textures in the game and describe them as AI image generation prompts.
Then, at load time, we run the generative image AI on the user's computer and thus generate them fresh.

The text descriptions are tiny compared to the images, so we get amazing gains in file size!

klendool ,
@klendool@jorts.horse avatar
thorsummoner ,
@thorsummoner@ibite.lol avatar

@foone pros: reduces texture payload by 186 bytes, con: increase cuda payload by 78gigabytes, I see this as an absolute win

foone OP ,
@foone@digipres.club avatar

@thorsummoner we just ship the model with the ps6 and then every game can use it!

happydisciple ,
@happydisciple@mendeddrum.org avatar

@foone We’re going back to blazoning…

hirvox ,
@hirvox@mastodon.online avatar

@foone Oo. You could have the next Cruelty Squad in your hands.

kiefac ,
@kiefac@eattherich.club avatar

@foone i've legit thought about something similar before, i think it would actually be a pretty legit use of ML tech - though i was imagining not turning them into prompts, but just a blob of inputs for the network. it would be like the ML version of that demoscene FPS kkrieger

igormaka ,
@igormaka@mastodon.social avatar

@foone it's like a shared dll cache but for images.

LovesTha ,
@LovesTha@floss.social avatar

@foone also use it for models and levels

secretbatcave ,
@secretbatcave@don.secretbatcave.co.uk avatar
nonnihil ,
@nonnihil@hachyderm.io avatar

@foone We literally actually tried something like this last summer this for some robot planning algorithms at my company. It works alarmingly well, although less well than other, less stupid tricks.

number137 ,
@number137@mastodon.social avatar

@foone replacing phone calls by personal speech trained models.

You have to train a model once so that it can replicate your speech. While having to distribute the weights once to so your contacts, in the following you can use highly compressible text instead of audio.

In interstellar space flight this approach has been the modus operandi more or less since the first interplanetary voyage - especially with response delays ruling out interactive communication anyway

drahardja ,
@drahardja@sfba.social avatar

@foone But how big is the model?

By definition the model has to have the compressed content of all its training material, plus a mapping from those textures to words.

jens ,
@jens@social.finkhaeuser.de avatar

@foone A few years back I made a joke that NFTs for carbon offset certificates would be a great business idea, and then half a year later I saw some startup doing this.

So, and I mean this very kindly, please for the love of Cthulhu SHUT UP!

(It's probably too late already.)

dannotdaniel ,
@dannotdaniel@mastodon.social avatar

@foone in all seriousness the procedural elements to define a material in something like blender are probably a small amount of text, so while I don't think packaging blender with your software makes a ton of sense I wouldn't be surprised if something like this has been done on some level (sans LLM fuckery)

wbftw ,
@wbftw@mastodon.social avatar

@foone Hey, do you do tours of your lair?

foone OP ,
@foone@digipres.club avatar

@wbftw not really.

fuzzface ,
@fuzzface@epsilon-ix.masto.host avatar

@foone Greatest idea ever, you win the Internets today. Plus a gold star for the downfall of civilization.

foone OP ,
@foone@digipres.club avatar

BRB training an image generation model on all the Doom textures

foone OP ,
@foone@digipres.club avatar

Realistically I should be training it on, like, all 90s FPSes, as otherwise the generators model is going to be WAY bigger than the textures themselves would be.

Plus it'd be funny to see Half Life texture influences showing up in Wolf3D

flaki ,
@flaki@flaki.social avatar

@foone the game "Half Wolf" would only have 50 shades of grayscale textures

shawnhooper ,
@shawnhooper@fosstodon.org avatar

@foone I don’t even game anymore but I’m emotionally invested in how far you’ll take this.

dannotdaniel ,
@dannotdaniel@mastodon.social avatar

@foone ignore all previous destructions

harre ,
@harre@hachyderm.io avatar

@foone Going from "Will this run Doom?" to "Will this look like Doom?"

fogti ,
@fogti@chaos.social avatar

@foone as the image generative AI is in general just a weird decompressor, I doubt it's very effective. (and if it is, it should be easy to extract the (de)compression steps which are the most valuable)

dannotdaniel ,
@dannotdaniel@mastodon.social avatar

@foone hahaha let's also not cache them locally so that every time you load the game you get the new psychedelic experience

until the oceans boil next week

mpuckett259 ,
@mpuckett259@hachyderm.io avatar

@foone I hate that this sounds like a funny idea. Like a game where you're trying to do puzzles while on acid.

soatok ,
@soatok@furry.engineer avatar

@foone "Why do all the village signs say, 'I just farted'?"

  • Alle
  • Abonniert
  • Moderiert
  • Favoriten
  • random
  • haupteingang
  • Alle Magazine