Wildest thing I've ever seen. And seems unwieldly too. But I really wonder how they are doing this.
Link: http://3DNes.com
Wildest thing I've ever seen. And seems unwieldly too. But I really wonder how they are doing this.
Link: http://3DNes.com
It's literally trying to figure out the depth from the rendered image (which is why it's so awfully inconsistent). There has already been the suggestion to just provide data separately for each game (much like how HDNes works if I recall correctly), at least that'd help making it stop flashing like that.
It looks like edge detection is used very badly somewhere in this, along with a depth calculation perhaps based on the draw order of the sprites?
It's a bit strange, and my eyes feel strange after trying to watch it
Games/Projects Currently In Development:
Discord RPG Bot | D++ - The Lightweight C++ Discord API Library | TriviaBot Discord Trivia Bot
It may be relying on palette indices by some point, to be fair (color 0 in the backgrounds seems to always be handled as max depth). Also I recall it actually processes the individual tiles to guess an intermediate depth that then gets used for the final frame computation. But yes, it's a horrible mess and honestly I'm surprised it works so well with Megaman o_O (it's like near perfect there) Also it seems to somewhat work in Contra until a player falls in the water and breaks its expectations of what's front and what's back (2D techniques don't mesh well with any attempts to translate them to 3D)
I wonder how it'd work if separate depth data for each tile was given to it instead of having to guess.