: As much as I want the human artist's hand to not be sullied, I see some
: practical applications for using an AI: - Procedural generation
: - (pretend there's a bunch of other bullet points under procedural
: generation)
: Like, take the automaps from games like doom and marathon. They do their job
: being a map for the player, but maybe an AI can take all those vertices
: and line segments fed to it from the map file, imagine how it should
: look, and make a stylized image of an area in the map using the player
: start as a vantage point? Or heck, stop having the task of making voxels,
: models, or 2.5d sprites drive me insane and have an AI figure out what a
: full set of sprites/full model should look like based on a couple images
: from a few key angles... I've thought on my own thought process going
: about it and how artists like Cheello make voxels... maybe methods and
: subroutines can be coded to imitate those artistic processes to make
: modding in new assets for 2.5D games like marathon less of a pain.
That would come in very handy indeed. Hopefully it can be used for more than just Marathon asset making, but GZDoom, Eduke32, Goldsrc, Source, Unreal & Tomb IDE/Tomb Engine assets too (to name a few of my favourite engines besides Aleph One). Quake engine stuff especially too as keyframes were a pain back in the day before skeletons.