There are so many things being tracked all the time in the game for puzzles and the power arm. Yet despites literally tracking sunshadows for some puzzle completion for example it runs almost smoothly with (in my 170h) no crashes. On a 6 yo portable console??
Botw was already impressive but I could grasp it with the shaders and also there weren’t that much physics puzzle. Objects were more static, there wasn’t the two other maps, enemy diversity was limited, same for weapons. There was less of everything overall but I thought it was the limit of the console and the possible engineering around it.
Is there any resources on how they managed to pull this off? White papers, behind the scenes, charts, …?
You should go read up on the small optimizations that developers do to seemingly surpass the limits of earlier hardware. Stuff like swapping palettes in between scanlines to give the impression of more available colors, or reusing palette swapped cloud sprites in place of bushes to save on limited ram. Good engineers are really, really good. They seem to find a solution for everything.
When I was extracting sounds from the sound banks from the Nintendo 64 game F-Zero X I simply couldn’t find certain sound effects. It turns out that some effects are created during runtime by taking a sound sample and applying certain effects or filters, for instance pitch shifting the sample and looping it in rapid succession.
It’s a clever way to save on memory and the player doesn’t notice if it’s well done. The original Pokémon Red/Blue on the gameboy is an example where it’s not so well done in some places. If you pay attention you’ll notice that some Pokémon’s battle cries are simple pitch shifts of other ones and they didn’t apply any other effects to obfuscate this.
I mean, pokemon needed 151 different battle cries. There is bound to be some overlap.
Pokemon is very optimised. In order to have the feature to give your pokemon nicknames, they had to scrap the other 105 pokemon just so they had enough space to write the code.
Sure, it’s a marvel how they managed to achieve a game of such scope within the constraints of a single cartridge. Still for the cries they could have managed to simply arrange them a bit smarter. For instance if you’re going to have pitch shifts the best place to use them would have been within evolutionary lines (grown up version simply has a lower pitch). So you could probably simply rearrange the existing sounds to make more sense without needing extra storage.
That’s exactly what I’m looking for ^^
Sorry, can’t give you references, but the examples I used were from one of the early Mario games. I was around when the euro demo scene was still hot, so there were loads of tips and tricks they used to game the 8086-x386 hardware to create dazzling effects in real time, prior to the introduction of acceleration and dedicated graphics chipsets. It was a truly glorious time, and a great source of wonder growing up in the 90s watching the industry evolve around me.
Still blows my mind (though not in a good way) to hear a simple app or http framework nowadays needs several gigabytess just to install. Everything I mentioned used to happen in 32k-8mb of RAM.
The demo scene still exists and they still produce mind blowing stuff.
I still remember “playing”
.kkrieger
, a 3d shooter fitting in a fucking 64kb exe. It did use quite a bit of RAM though.Ah yes! Projecktt or something like that. Procedurally generated models and textures, and caused some controversy due to heavy reliance on the directx API. But technically astounding, nonetheless!
Produkkt