POLYGON VIDEO GAMES: A VISUAL HISTORY
Every time I boot up a new engine and drop a cube into the scene, I'm doing the same thing someone at Atari did in 1979 with a room full of vector hardware. The triangle is the oldest trick in 3D graphics. What changes is how many of them we can afford, and what we're willing to trade to push the count higher. The whole history of polygon video games is basically a four decade argument about that tradeoff.
I wanted to write this because the word "low poly" gets thrown around like it's one thing, a single aesthetic that showed up around 2014 on Steam. It isn't. Low poly is the ghost of every hardware limitation developers ever worked around, stacked on top of each other. When I build something chunky and faceted today, I'm borrowing from PS1 texture warping, N64 vertex snapping, Virtua Fighter flat shading, and Battlezone wireframes all at once. Understanding where those looks came from makes you way better at using them on purpose.
So here's the tour. Every era of polygon games, what the hardware could actually do, and the decisions that got baked into what we now think of as the polygon look.
1980, Battlezone and the wireframe era
Battlezone didn't run on a raster screen. It used vector graphics, which means the electron gun in the CRT traced lines directly instead of scanning rows of pixels. That's why the geometry looks so clean. There's no rasterization aliasing because nothing is being rasterized. The screen is drawing the polygons as actual lines.
The tank, the pyramids, the distant volcano, all of it is wireframe. Hollow polygon shells with no fill. You can see through every object. The game ran on custom vector hardware, the same stuff powering Asimov and Tempest and Lunar Lander, and the polygon budget was tiny. A few dozen line segments on screen at a time, max.
What's wild is how modern it feels. Grid floor, monochrome green, distant mountains, first person vehicle. If you squint, Battlezone is the same composition as half the synthwave indie games on Steam in 2018. The look came back because the constraints make sense. Wireframe reads instantly, even on bad monitors, even from across an arcade.
Atari followed up with Red Baron and I, Robot, the latter being the first commercial game with filled polygons instead of wireframes. 1984. Flat shaded 3D in a cabinet with a special processor just for matrix math. Nobody bought it because the gameplay was weird, but the engine was a decade ahead of what home computers could do.
1983, Star Wars arcade and the vector high point
Star Wars came out three years after Battlezone and pushed vector hardware to its limit. The Death Star trench run, TIE fighters chasing you, Darth Vader's voice hissing from the cabinet, all rendered in glowing colored lines. Because vector screens can push individual line brightness way past what a raster CRT could manage, the whole thing has a specific neon quality that nothing since has quite matched.
I've played the cabinet a few times and the thing that always gets me is how the polygons feel like they're physically inside the screen. The phosphor glow, the slightly shifting lines, the way fast moving objects leave a tiny trail of afterimage. It's analog 3D. None of it is simulated. The electron gun is literally drawing the TIE fighter in real time.
This era ended fast once filled polygons and sprite scaling got cheap. Vector hardware was a dead end commercially because you couldn't do textured surfaces or detailed characters. But the look, that hard edged neon wireframe, lives on in every game that wants to feel retro futuristic without going full PS1.
1993, Virtua Fighter and flat shading takes over
Sega's Model 1 board was the first arcade hardware built to do real time textured and flat shaded 3D at 60 FPS. Virtua Fighter showed up in 1993 and changed everything. The characters were around 1,200 polygons each, flat shaded, no textures, and the whole thing ran at a locked frame rate.
If you pull up screenshots now, Virtua Fighter looks goofy. Sarah Bryant's face is a few triangles. Hair is literally a single angular chunk. Jacky has hands shaped like boxing gloves because that's what the polygon budget allowed. But it moved at 60 FPS with proper animation and readable hit detection, and that combination of motion and form is why people went crazy for it.
This was the moment when polygon video games started eating sprite games. Virtua Racing, Daytona USA, Virtua Cop, all on Model 1 or Model 2. Arcades became polygon machines. The flat shaded look, where every face was a single color computed from the light direction, defined the era because textures were still expensive. You got your aesthetic from color choice and geometry, not surface detail.
I still think flat shaded polygons are the cleanest way to render a character. There's no noise, no texture stretching, no resolution dependency. The shape does all the work. When I look at modern stuff like Sludge Life or Crow Country, I see the Virtua Fighter DNA in there, just with more polygons and better cameras.
1996, Super Mario 64 and the N64 weirdness
Mario 64 is the game every polygon history article wants to talk about, and fair enough. It was the first 3D platformer that actually controlled well, and it defined 3D level design for a decade. But the visual side of it is weirder than people remember.
The N64 had a very specific set of tradeoffs. It could do bilinear texture filtering, which was rare at the time, so textures looked smoother than on PlayStation. But it had a tiny texture cache, only 4KB, which meant textures had to be small and repeated. That's why N64 games look blurry and muddy. The smoothing filter plus postage stamp textures stretched across big surfaces.
The polygon counts were also low. Mario himself was around 750 triangles. The levels were simple chunks of geometry with Gouraud shading and vertex colors. Peach's castle, Bob-omb Battlefield, the whole visual identity came from color palette, geometry silhouettes, and music, because there was no texture detail to carry the mood.
Here's the thing. That look wasn't chosen. It was what the hardware could do. Nintendo made the best of a constrained texture pipeline by doubling down on chunky readable geometry and bold color. And because it worked, a whole generation of kids grew up thinking that's what 3D was supposed to look like. Which is why N64 nostalgia is a specific flavor of low poly that doesn't match PS1 nostalgia at all, even though both were 32/64 bit era.
1994 to 1999, PlayStation and the texture warping look
The PS1 did polygons differently than the N64. It had more texture memory but no filtering, so textures stayed sharp and pixelated. No floating point on the GPU either, which meant vertex positions were snapped to integer coordinates every frame. That's where the famous PS1 wobble comes from. Polygons jiggle because their corners are being rounded to the nearest whole screen pixel.
There's also affine texture mapping, which means the textures didn't account for perspective correctly. Walls and floors look like they're sliding and warping as the camera moves, especially on big polygons close to the camera. Combine that with the vertex snapping and you get the unmistakable PS1 texture dance. Metal Gear Solid, Silent Hill, Tomb Raider, Final Fantasy 7 battle scenes. All that chaotic surface motion is the GPU struggling to do math without a floating point unit.
This is the era that the PSX horror revival crowd is chasing, and it's wild to me because the thing they're replicating is literally a hardware bug. The warp isn't a style choice from Konami. It's a math limitation they couldn't fix. And now it's an aesthetic. I wrote a bit about this kind of reclamation in my piece on what low poly actually means if you want to go further down that rabbit hole.
Character polycounts on PS1 were usually around 300 to 800 triangles. Cloud in FF7 is famously around 170 triangles for his overworld model, with most of the detail in textures. It's astonishing how much personality those chunky textured shapes carry. Pre rendered backgrounds, full motion video cutscenes, everything was optimized around the 2MB of RAM and the 33 MHz CPU. Constraints breed style.
2000 to 2005, the sixth generation and mid poly
The PS2, GameCube, Xbox, and Dreamcast before them cranked polygon counts up by roughly an order of magnitude. Characters went from hundreds of triangles to thousands. Textures got bigger. Bump mapping started showing up. This era is less iconic visually because it's the awkward middle. Games weren't chunky enough to be stylized, but the hardware still forced a ton of shortcuts that make the visuals look dated now.
Shadow of the Colossus, Wind Waker, Jet Set Radio Future, these are the games from this era that still look great. Why? Because they all committed to a visual style instead of chasing realism. Wind Waker went full cel shaded. Colossus used atmospheric fog and motion blur to hide polygon density issues. Jet Set was hard inked outlines and flat color. The games that aimed for realism in this era aged badly. The games that used the polygon budget for style held up.
This is a lesson I think about constantly. Polygon budgets don't determine how a game looks forever, style choices do. A 500 triangle character with good silhouette and color beats a 5,000 triangle character with muddy textures every single time.
2006 to 2013, normal maps and the HD stalemate
The 360 and PS3 brought normal mapping into the mainstream. Suddenly you could have a 20,000 triangle character with a baked normal map that made it look like a million polygons worth of surface detail. Gears of War, Uncharted, Battlefield, all built on this technique. The polygon budget question got temporarily buried under "just bake more detail into textures."
For most of this generation, games got shinier and browner and more realistic and kind of visually boring. The polygon pipeline hit a plateau where more polygons didn't really matter because normal maps were doing the surface work. It took a while for anyone to get interesting again.
Then indie 3D started showing up. Proteus, Fez, Bastion, Monument Valley. Small teams making stylized polygon games that didn't care about photorealism. The low poly revival as a deliberate movement starts here, because indies realized that if they couldn't compete on texture density and shader complexity, they could compete on pure style. Flat shaded, chunky, hand colored, whatever. The polygon became a design choice again instead of a limitation.
2014 onward, the low poly revival and the triangle as icon
Somewhere around 2014, low poly stopped being retro and started being its own thing. Super Hot came out in 2016. Mini Metro, A Short Hike, Untitled Goose Game, Islanders, there's a huge cluster of games now that use simple polygon forms as their entire visual identity.
What's different from the Virtua Fighter era is that these games aren't constrained. A 4K render of a modern low poly game is pushing way less geometry than the hardware could handle. The low count is a choice. The flat shading is a choice. The hard edges are a choice. Every aesthetic decision is deliberate, which changes what the look means. It's not making the best of limitations, it's using limitations as a design vocabulary.
I've written a lot about the specific craft of this, including the math behind why simple polygon forms are so visually pleasing in the geometry behind poly art. The short version is that the human visual system is really good at reading silhouettes, and polygon art amplifies silhouette by stripping everything else out.
2020 onward, Nanite and the end of polygon budgets
Unreal Engine 5 shipped Nanite in 2021 and it changed the rules again. Nanite is a virtualized geometry system that can render scenes with literal billions of triangles by streaming in geometry at sub pixel detail. The polygon budget, the single biggest constraint in 3D graphics for forty years, basically went away for teams using this pipeline.
What does that mean for the polygon look? Honestly, it makes the stylized stuff more interesting. If any indie studio can push photorealistic geometry now, then choosing to make a chunky flat shaded game is an even stronger artistic statement. The constraint is gone, so the decision to impose your own constraint becomes the whole point of the style.
It also means the old techniques don't disappear, they just become tools. You can mix Nanite environments with hand built low poly hero characters. You can use PS1 style texture warping as a deliberate shader effect on top of a modern lighting setup. The whole history of polygon rendering is sitting there as a library of looks to pull from.
Why this history matters if you're making stuff
Every polygon game I play, I try to figure out which era it's drawing from. Is this N64 chunky with blurry textures, or PS1 chunky with sharp textures? Is it flat shaded like Virtua Fighter or Gouraud shaded with vertex colors? Is it using modern shader tricks to fake an older style, or is it actually built with low poly meshes?
The answers matter because each of those choices came with baggage. PS1 style carries horror and memory and grit. N64 style carries platformer bounce and childhood. Flat shaded Model 1 style carries arcade snappiness. Battlezone wireframe carries sci fi minimalism. If you pick a look without knowing where it came from, you'll accidentally import vibes you didn't want.
When I start a new project now, I sketch out which era I'm stealing from on purpose and which tricks I'm leaving behind. Last thing I shipped pulled silhouette work from Virtua Fighter, vertex colors from the PS1 era, and modern shader anti aliasing from the 2020s. Mixing eras on purpose is way more powerful than picking one and copying it.
The polygon is still the same building block it was in 1980. What we've got now is forty years of ways to use it. Pick the ones that fit your game, ignore the rest, and keep making weird angular things. That's the whole trick.
LIKED THIS? STAY IN THE LOOP
New posts, game updates, and things you won't find anywhere else.