The problem with 1985
Real-time polygonal 3D in 1985 was not a matter of ambition. The arithmetic required to transform, project, and fill textured polygons at sixty frames per second was well beyond what arcade hardware could do at any price a cabinet could bear. Games that wanted depth had two options: fake it with a scrolling road and a few sprite sizes, as most driving games did, or accept wireframes, as Atari's vector games did.
Yu Suzuki and Sega's AM2 division found a third route. The Super Scaler boards, introduced with Hang-On in 1985, calculated positions, scales, and zoom rates in three dimensions — genuine 3D maths, running on dedicated hardware — and then converted the results back into two-dimensional display data. The machine knew where everything was in space. It simply drew the answer as a sprite at the correct size rather than as geometry.
Scaling as texture mapping
The insight is sharper than it first appears. Scaling a sprite according to its computed distance is doing, in a restricted form, what texture mapping would later do on polygonal hardware: taking a two-dimensional image and presenting it at a size and position determined by three-dimensional position. Sega's boards handled scaled sprites and tiles in a manner directly analogous to how the texture-mapped 3D games of the 1990s handled textures. The technique was not a hack that happened to look 3D; it was an implementation of the same underlying idea, restricted to objects that always face the camera.
That restriction is the entire cost. A scaled sprite cannot be viewed from another angle — it has no other angle — so Super Scaler games are built around forward motion, where everything you see is coming toward you and rotation is limited. Hang-On, Space Harrier, After Burner, and OutRun are all games about hurtling forward, and that is not a stylistic preference. It is the shape of the games you can build when your objects have exactly one face.
Sixty frames, no slowdown
The performance is what actually made these machines famous. Space Harrier's hardware included an Intel 8751 microcontroller running at 8 MHz and displayed a claimed 32,000 colours; where Hang-On had suffered occasional framerate dips and sprite flicker, Space Harrier ran at a constant sixty frames per second with multiple enemies at different scales, obstacles, bullets, and a continuously scrolling background, without detectable slowdown.
Sixty frames per second, in 1985, with dozens of objects scaling smoothly toward the player. Home hardware would not touch that for a decade, and the contemporary machines that were attempting genuine 3D — vector cabinets, early polygon experiments — were nowhere near it. A Super Scaler cabinet in an arcade was doing something visibly impossible, and the visible impossibility was the product. Sega was selling an experience the customer could not have anywhere else, which is exactly what a coin-op business needs to sell.
The bridge
The Super Scaler era ended because Suzuki himself ended it. Having spent the late 1980s extracting the maximum from fake 3D, AM2 moved to the real thing with Virtua Racing in 1992 and Virtua Fighter in 1993, and the sprites went away. It is a strange career shape — the person who perfected the workaround was also the person who made it obsolete — and it suggests Suzuki had never been attached to the technique, only to the effect it produced.
What the Super Scaler years demonstrate is that the gap between what hardware can compute and what it can draw is exploitable. Sega's boards were doing 3D mathematics years before they could render 3D geometry, and Suzuki noticed that the maths was the valuable part — knowing where things are in space is what makes a game feel three-dimensional, and the drawing is negotiable. That is the same insight that later underpins every rendering shortcut in the business, from billboarded particles to impostor sprites in modern engines. Super Scaler was not a stepping stone toward real 3D. It was the recognition that "real" was the wrong question.