by basdp on 11/8/16, 8:49 PM with 7 comments
by camtarn on 11/8/16, 10:25 PM
The 3DNes emulator doesn't just automagically convert 2D games into 3D. It's got some algorithms behind it that attempt to partition a given scene into regions, then assign the region to a layer (front/back) and extrusion method (box, horizontal/vertical cylinder, character). Someone then needs to play through the game, and if they see anything that the algorithm has guessed wrong, they can pull up an editor window, click on the region in the emulator, and edit layer/extrusion/etc on the fly. Once that's all done, the collected data can be uploaded so that other people playing the same game can play it with everything looking correct.
I'm not entirely clear on whether the partitioning/assignment/settings memorization works by detecting the actual sprites or graphics tiles that the emulated NES is drawing to the screen, or whether it's just calculating some sort of signature for that clump of pixels and looking it up in a table. My guess is the latter.
It's a very nifty piece of tech :)
by voltagex_ on 11/8/16, 9:47 PM
by TazeTSchnitzel on 11/8/16, 9:29 PM
by gravypod on 11/8/16, 10:12 PM
by planteen on 11/9/16, 6:08 AM