The room you're in is a reconstruction of an early 3D surfacing program that shipped in 1985 for a UNIX workstation costing more than a small house. This is not that workstation. This is a browser, reaching back forty-one years to imitate a thing it was not present for.
We have no ROM, no original assets, no fragment of the source. What we have are surviving screenshots, papers from the SIGGRAPH archive, the founders' memories recorded much later. The chrome you see is reconstructed. Cardinal splines really do pass through their control points — that part is honest. Revolve really revolves and Sweep really sweeps. Everything else is polite forgery, signed and dated.
What works in this room. Drag a yellow CV in the FRONT viewport — the profile curve and the live surface rebuild in real time. Use the palette's +CV / −CV to add or remove control points. Move, Rot, and Scl apply one-shot transforms to the whole profile around its centroid (click repeatedly to compound). Toggle between Revolve (around Y axis, 360°) and Sweep (gently arced path along Z) with the buttons under the toolbar, or via the Surfaces menu. Change tessellation in Surfaces > Segments. Reset the demo profile from Edit > Reset Profile. Use the Rooms menu at the top-right to jump to another cartridge.
Stephen Bingham, Nigel McGrath, Susan McKenna, David Springer. Four founders, one Toronto. The company is named after the anti-aliasing technique their early renderer used — a small inside joke for the rendering crowd.
Two years before Alias/1 ships, they are writing the code that will revolve the curves that will render the surfaces that will, eventually, in a few films you have heard of, be a dinosaur. That is later. Right now it is Toronto, 1983, and the math is the work.
Jim Clark and Marc Hannah, at Stanford from 1979 and demoed in 1981, before Silicon Graphics existed, designed a VLSI chip that did one thing extremely well — multiply a 4×4 matrix by a vector, in hardware, fast. The Geometry Engine.
SGI was the company built around the chip. Alias/1 was the application built around the workstation built around the chip. Every dinosaur you remember is somewhere in this lineage of multiplications happening in silicon while you watched, frame by frame, polygon by polygon.
The Geometry Engine drew the wireframe fast — but a wireframe is ambiguous, near edges and far edges tangled in the same ink. Hidden-line removal is what turns it into a readable solid, and it was Alias's killer marketing demo (toggle HLR in the perspective viewport to watch it). The technique is older than SGI by two decades: Lawrence Roberts solved it analytically in his 1963 MIT thesis, testing each silhouette edge against every polygon that might hide it — and that analytic approach was still the professional standard when Alias shipped in 1985. This room cheats: it fakes the same look with a modern depth-buffer pass rather than Roberts's edge math, and tells you so rather than letting the forgery pass for the original.
1985. The first SGI workstation with a window manager — MEX, Multiple EXposure, a proprietary thing that ran on SGI's own UNIX variant. 68010 CPU at ten megahertz. Up to eight bit-planes of color. 1024 by 1024 pixels at thirty hertz interlaced.
The display alone weighed as much as a small adult. Forty-five thousand dollars on the low end. This is the box Alias/1 ran on, and the chrome you are reading this through is what it might have looked like, faithfully or not, through that bit-mapped phosphor.
(Display correction, 2026: earlier drafts of this room said 1024×1024 at 30Hz interlaced — that was the IRIS 1000/1200, not the 2400. The 2400's panel was 1024×768.)
A NURBS curve does not pass through its control points — it is pulled toward them but stays separate. A Cardinal spline does pass through its control points, with a tension parameter controlling how tightly it hugs them.
Alias/1 chose Cardinal. The math was simpler, the GL hardware did Cardinal-friendly operations natively, the interaction was more direct. Move a point, the curve goes there. The later switch to NURBS was a generality move; in 1985 the simpler curve was the right one. Drag a yellow CV in the FRONT viewport and you can feel the difference — the curve obeys you rather than negotiating with you.
Alias/1 was not a modeler. It was a modeler, a paint program, an animation system, a renderer, and a film recorder, all in one application — a thing almost no software does anymore. Today you buy six packages and stitch them. In 1985 you bought one, and the integration was the product.
The paint system, in particular, was the first paint program ever shipped for that workstation. Note the pattern — the suite-inside-the-app, where each tool feeds the next. Classicery is doing something a little similar, four decades later, with different reasons.
Inside Alias/1, the paint system painted directly onto rendered surfaces. You revolved a hull, you painted on it, you re-rendered. Standard now — every 3D app has texture painting — but in 1985 it was new, and the loop from model to paint to render to film was tight enough to actually use.
We have not implemented the paint system in this vitrine. We have a Revolve tool and a Cardinal spline editor. The paint would have been a third room inside this room, and this is already an ambitious museum exhibit.
Alias/1 in 1985. Alias/2 by 1987. PowerAnimator from 1988 through 1999 — the SGI workstation era, the Jurassic Park dinosaurs, the merger with Wavefront in 1995.
Then Maya 1.0 in 1998, the rewrite that killed PowerAnimator and reset the field. Then a long period at Autodesk after the 2006 acquisition. Alias still exists today as Autodesk Alias, the Class A automotive surfacing tool. Every car you see on the road was finished, at some point, in something descended from that four-founder Toronto application.
The integrated-suite idea did not survive the unbundling that followed. The Cardinal spline lost to NURBS. The MEX window manager was gone by 1988. The IRIS 2400 became a museum piece — and now, recursively, an exhibit in another museum.
What stayed is the lineage. The founding gesture of putting modeling and animation and paint and rendering and film recording into one application, the belief that 3D should be an integrated discipline, the Toronto company that became the company that became part of Autodesk. And the curve. Some version of the curve is still there, in every NURBS surface every industrial designer touches.
Two sibling 3D rooms down the hall: SV (cart 013, MAGI SynthaVision 1972, CSG ray casting — solids as boolean expressions) and BLD (cart 012, Cambridge BUILD 1973, B-rep solid modeling — solids as bounded surfaces). Both took the volume-and-Boolean path Alias declined. Spline surfaces close form; boolean expressions and B-rep boundaries close volume. Today's CAD contains all three.
The room you can build in: POINTS — the working fork of this museum room. Cubic Bézier curves, revolve / extrude / sweep, quad-mesh OBJ export. For very beginner design students. The math is modern; the chrome stays close to what you see here.
If you want to work in the NURBS register today, the other room next door is the Plug Innery — a small suite of plugins for Rhino, the contemporary descendant of the Alias surfacing approach. Different generation, same curve.
From THE LITERATURE · CANON / 001B · the theory wing of CLASSICERY. Hear Edwin Catmull and Jim Clark on Recursively Generated B-Spline Surfaces on Arbitrary Topological Meshes read aloud — → track 09 →
→ CANON3 · track 06 · Kochanek–Bartels splines · the curve handed back
→ CANON5 · track 10 · Nervous System · surfaces grown, not drawn