by 
The original trio of Pokémon games for the Game Boy are now playable in a beautiful new format, thanks to two open-source projects that received major updates today.
The first project is gen1recomp, a complete rewrite of the original game engine that takes a standard first-gen Pokémon ROM as input but uses no Nintendo code whatsoever - allowing for complete playthroughs of Pokémon Red, Blue or Yellow while adding modern features like higher refresh rates, game speeds and mod support.
That extensibility is used to full effect with the DramaticShapeVoxelMod, which interprets the original game data to produce a beautiful voxelised world, from interior and exterior scenes to full battle sequences using the original sprites.
The combination has to be seen to be fully appreciated, and thankfully the two GitHub repos listed above are relatively straightforward. After running the latest gen1recomp release for your system, you can provide a legally obtained Red, Blue or Yellow US ROM to the program, then visit the mods tab and select the voxel mod. From there, you can boot up the game, enable the various features added by the mod, and explore a world of dreams and adventures with Pokémon.
Playing through just to the first gym battle was an absolute treat, and despite the fact that these are still early releases, I didn't encounter any game-breaking bugs. There were only a couple of oddities too, such as the old man at the north of Viridian City having placeholder text, battle transitions not extending to fit the full screen and the voxel graphics flicking into place momentarily when entering a new zone. Overall, it's quite a smooth experience.
However, there is some debate over the recompilation project's origins, stating that the code shows signs of AI generation despite a lack of public disclosure. That would be a real shame, as it would limit how polished the project could become in the future.
Do you like the look of the mod's tilt-shift graphics, or do you prefer the original flat look - or even the modern remakes? And where do you stand on the AI coding front for projects like these? Let me know in the comments below.