Gotta say, for a 90s console game on 16-bit technology, it looked phenomenal. Both this and Super Mario RPG.
Rare used top of the line super computers for the CGI graphics, SGI machines. The same kinda computers used in the making of Jurassic Park, iirc.
I might be mistaken, but didn’t they use the SNES Super FX chip in this game? I know it was used in Star Fox.
Pre-rendered graphics.
They rendered each sprite from a 3D model on a supercomputer, then copied that image to the SNES ROM. All the SNES did was load the readymade images as needed. No SuperFX chip, but the original DKC carts had more ROM storage than any SNES game before it.
And then Rare claimed they were continuing the memory-maxing shenanigans with DK 64 requiring the Expansion Pack to play, but it was really only to fix one bug that would be hard to run into but would break the game without it. The Expansion Pack does very little to nothing for the game otherwise. According to some former Rare devs they could’ve fixed the bug without needing the Expansion Pack but they decided to require it anyway.
That ended up being a myth. It really does use that extra RAM. https://www.nintendolife.com/news/2019/11/feature_donkey_kong_64_devs_on_bugs_boxing_and_20_years_of_the_dk_rap
Nah, it’d have been overkill for sprites. The sprites were based on models built using insanely expensive work stations, digitized into 2D sprites to fit on a SNES cart.
Counterpoint, 1994 also gave us this graphical masterpiece
Well, the game looked close, but not like that
Dehance
Best graphics in 1994 on a technical level would probably be something like Virtua Fighter 2. But almost all early polygon games have aged like ass, unlike sprite-based ones.
Not sure I’d call that best on a technical level. It was doing something different: projecting triangles from 3d space to 2d space in real-time.
Donkey Kong country also did that but just not in real time. It was 3d graphics but pre-rendered and used as sprites. The rendering process would have been at least as technical.
Mortal Kombat was also around the same time and used sprites that were based on live action captures and involved highly technical stuff in a different direction.
Though if I had to guess, Doom was probably the most technical at the time, since it did 3d rendering in real time without relying on any 3d-specific hardware.
Iirc the 3dfx chip in some snes games handled that for games like starfox and the PlayStation had a hardware 3d renderer pipeline. Neither were particularly powerful (which is why early 3d stuff looked like cyber trucks, because polygon count needed to be kept low to hit frame deadlines), but they did offload that from the CPU.
For Doom, it was all handled on the CPU, which is why it can run on pretty much anything with a CPU and a display. Carmack figured out a lot of cool optimization tricks, like a fast square root approximation, to make it possible.
The fast square root didn’t come along until Quake 3.
Doom was kinda 2.5d, not 3. They swapped out 2.5 d levels when you entered an area where you supposedly overlapped the map in the vertical axis. Other games that came out around the same time include worlfenstein 3d (also 2.5d, plain floor iirc), ultima underworld (not full screen) and system shock (full 3d, but always oriented to the user, with feet on the ground). Only descent went full 3d, but was disorienting to some users. Dooms tour de force was the optimization and the fps/ immersion it achieved, and we have carmack and his work on raycasting to thank for that, on (you stated) non accelerated hardware.
They still elite ngl
Also made on super expensive computers. That cost around $25k a piece.
I think there’s something charming about like Rez graphics in games. I don’t personally need photorealism.
My hot take: yes, the Donkey Kong Country graphics looked very impressive at the time, as a technological feat that gave a 3D-rendered appearance on a console not capable of 3D rendering*.
But that doesn’t mean they were actually good.
Nowadays, I think it’s pretty well accepted that rendering your 3D model into sprites is the quick/lazy way to do it, and generally produces inferior results to creating your pixel art by hand to get things like the dithering and appearance after CRT blur is applied just right. Praising these sprites for being 3D-rendered is looking at them a bit with rose-tinted glasses.
More importantly, while Donkey Kong Country may have looked pretty at first glance, IMO it didn’t play as well as a good pixel art platformer like Super Mario World. A big part of the reason for that was that the imprecision of the sprites (leaving the edges at whatever pixel the floating-point-based 3D renderer happened to round them to, as well as in some cases obscuring those edges in shadow) made it harder to tell exactly where the platforms ended and where the character was in relation to them.
(* other than flat-shaded triangles with a SuperFX chip, which is not even close to the complexity of the DKC renders)
