There is a version of this project where I could simplify everything. Flatten the canopies into single-layer plates. Skip the bleacher seating. Ignore the angled facade glass. Build something that reads as "Pagoda-shaped" from across a room and call it done. That version would be faster, cheaper, and easier to engineer. It would also miss the entire point.
The IMS Pagoda is not a simple building. It has canopy overhangs that appear to float in midair. It has glass facades that slope inward at angles LEGO was never designed to replicate. It has bleacher seating integrated into the ground floor structure, a rooftop observation deck with railings, elevator shafts visible through the glass, interior corridors, and individual rooms on every floor. At 1:38 scale, all of that detail is either captured or the model is just a shaped stack of bricks. I chose to capture it.
That choice has consequences. Every detail that goes into this build introduces a design problem that has to be solved within the constraints of LEGO geometry. Bricks connect at right angles. Plates come in fixed increments. Slopes exist at specific pitches. The real Pagoda does not care about any of those limitations. So every floor of this model is a negotiation between what the building actually looks like and what LEGO parts can physically achieve.
This article covers the major detail challenges — some solved, some still open — and the design philosophy that guides every decision.
Every floor of the IMS Pagoda features canopy overhangs that extend outward from the building face, shading the windows below and giving the structure its distinctive layered, horizontal profile. These overhangs are not decorative afterthoughts — they define the building's silhouette. Get them wrong and the entire model reads wrong.
In the LEGO model, each canopy extends approximately 10 to 12 studs from the building face. At 1:38 scale, that translates to roughly 10 to 12 feet of real-world overhang — consistent with what the reference photos show. The structural challenge is straightforward: cantilevering plates outward and supporting them from below without visible bulk. Technic pins, plate-on-plate layering, and bracket attachments handle the structural side well enough.
The aesthetic challenge is the real problem. On the actual Pagoda, the leading edge of each canopy has a thin, almost translucent quality. The canopy tapers to something that looks like glass or a very thin metal edge. In photographs, especially at sunrise and sunset, the edges catch light and appear to glow. It is one of the most distinctive visual features of the building.
Replicating that in LEGO has proven to be the single most difficult design problem in the entire project. I have investigated multiple approaches:
Trans-clear curved slopes (#11477 Slope Curved 2×1) — These are the obvious first choice. A transparent curved slope at the edge of each canopy would suggest the glass-like taper. In practice, the curvature is too pronounced for the scale. The edge reads as "rounded" rather than "thin," and the trans-clear plastic does not quite capture the frosted, light-catching quality of the real canopy edges.
Light Aqua parts — LEGO's Light Aqua color has a pale, watery quality that, in renders, approximates frosted glass. But it is opaque, not translucent. Under certain lighting it works; under others it reads as a solid color block rather than a transparent edge. It also introduces a color that appears nowhere else on the building, which disrupts visual consistency.
Bright Light Blue parts — Similar problem. The color is close to what the eye expects for tinted glass, but the opacity kills the illusion. A Bright Light Blue edge piece on a white canopy looks like someone painted the edge blue, not like light passing through glass.
The canopy edge treatment remains the biggest unsolved design problem in this project. Every option I've tested is a compromise. I haven't found the right part yet — and I may not. Sometimes LEGO simply cannot do what a real building does.
For now, the design uses a placeholder approach: a single-plate-thick edge with no special transparency treatment. It looks clean. It looks intentional. But it does not look like the real building's canopy edges, and that gap between the model and reality is something I think about every time I open the Stud.io file.
The IMS Pagoda has glass facades that slope inward — the building's face is not a simple vertical wall. The windows tilt at an angle that gives the structure its aggressive, forward-leaning appearance. It is subtle enough that you might not notice it in a casual photograph, but the moment you try to build a straight vertical wall in LEGO and compare it to the real building, the difference is immediately obvious.
LEGO does not do arbitrary angles. You get 90 degrees. You get specific slope pitches — 33 degrees, 45 degrees, 65 degrees, 75 degrees. The real Pagoda's facade angle falls somewhere between standard LEGO slope increments, which means any solution is an approximation.
The parts under investigation are #2449 (Slope Inverted 75° 2×1×3) and its corner counterpart #2468. These inverted slopes mount on the underside of a plate and angle outward at 75 degrees from horizontal — or equivalently, 15 degrees off vertical. The real building's facade appears to tilt at something close to 10 to 12 degrees off vertical based on photo analysis, so 15 degrees is close but not exact.
The question is whether "close but not exact" is good enough at this scale. At 1:38, a few degrees of angular error translates to fractions of a millimeter at the stud level. The eye is remarkably forgiving when the overall proportions are correct. My current assessment is that #2449 and #2468 will work — but I will not commit until I have tested them physically. Stud.io renders can be misleading when it comes to subtle angles, because the rendering engine smooths geometry in ways that real plastic does not.
There is also the connection problem. Inverted slopes need to attach to something, and that something needs to be structurally integrated with the floor plate above and the window assembly below. Every angled facade element adds complexity to the connection sequence for that floor module. In a modular build where every floor needs to separate cleanly, adding connection complexity is a cost that has to be weighed carefully.
The base of the IMS Pagoda includes a bleacher section — tiered seating that faces the track. This is one of the most recognizable features of the ground-level structure, and capturing it accurately was a priority from the early design phase.
The bleacher design uses four rows of seating, each stepping up from front to back. The section is 19 studs deep in total, with the four rows sitting at heights of 3, 5, 7, and 9 bricks respectively. Between each row, a 2-stud-wide walkway provides the visual separation that makes the individual seating levels readable.
The seating itself is built in black — consistent with the real Pagoda's bleacher appearance. At 1:38 scale, each row of seating is narrow enough that individual seat detail is not possible. Instead, the stepped profile and color contrast against the surrounding structure is what sells the illusion. From a few feet away, the model reads unmistakably as tiered seating, which is the goal.
The glass facade in front of the bleachers is another detail that required careful thought. The real building has tall glass panels that extend upward from the ground level, partially enclosing the bleacher area. In the model, transparent panels and glass-effect elements recreate this enclosure. The challenge is maintaining structural rigidity — transparent LEGO elements tend to have looser clutch tolerances than opaque bricks, so a wall of trans-clear panels can feel fragile. The solution involves sandwiching transparent panels between opaque structural elements at regular intervals, creating a wall that looks like continuous glass but is actually segmented for strength.
The Pagoda's rooftop features an observation deck with railings — a distinctive cap to the building's profile. For the model, the railings use part #3185 — the Fence 1×4×2 Lattice in black. This part has the right scale, the right visual density, and the right color. It reads as a metal safety railing at 1:38 scale without being so visually heavy that it overwhelms the roofline. The lattice pattern also allows light through, which maintains the airy, open quality that the real observation deck has.
Below the rooftop, the interior details are more ambitious than most MOC builders would attempt at this scale. The Stud.io renders reveal elevator shafts running vertically through the building, individual rooms on each floor, and corridor layouts that connect the functional spaces. None of this interior detail will be visible in the finished, assembled model — but it is visible when individual floors are removed, which is one of the advantages of the modular design approach.
The elevator shafts are particularly important. The real Pagoda has clearly visible elevator infrastructure on the exterior — you can see the shaft housing from outside the building. In the model, the elevator section uses a combination of transparent panels (for the glass enclosure) and gray structural elements (for the shaft framing). The result is a section that looks mechanically purposeful rather than decoratively hollow.
Each floor includes enough interior detail that if you were to cut the model in half, it would look like a building, not a box. Walls divide rooms. Corridors run between them. Stairwells connect floors at consistent locations. This level of interior work adds significant design time to each floor module, but it also means the model holds up to close inspection from any angle — including straight down through a removed rooftop.
Every detail in this build exists in tension with LEGO's geometric constraints. The real building has continuous curves, arbitrary angles, smooth gradients, and surfaces that transition seamlessly between materials. LEGO has studs on an 8mm grid, plates in 3.2mm increments, and slopes at fixed pitches. The gap between those two realities is where this project lives.
The approach I have settled on is what I think of as honest approximation. If LEGO cannot replicate a detail exactly, I would rather use the closest available part and let it be visibly LEGO than try to hide the medium. The model should look like a LEGO building — not a building that happens to be made of LEGO. The stud texture, the plate lines, the slope transitions — these are features, not flaws. They are what makes this a LEGO model rather than a 3D print or a scale model kit.
That said, honest approximation still demands precision. The canopy overhangs must be the right depth even if the edge treatment is imperfect. The bleacher rows must be the right count and the right relative heights. The facade angle must be as close as the available slope parts allow. Approximation is not an excuse for laziness — it is a design philosophy that acknowledges the medium while still demanding accuracy within its constraints.
One of the most important rules I have established for this project is simple: never guess a part number. Always verify on BrickLink before committing a part to the design.
This rule exists because I broke it. Multiple times. Early in the design process, I referenced parts by numbers I thought I remembered or that seemed logical based on naming conventions. Part #34146 appeared in my notes as a specific bracket type. It does not exist. I had fabricated the number from a misremembered search result. Part #62113 was listed in an early version of the ground floor design as a specific panel element. It turned out to be a completely different part than what I intended — a misidentification that would have resulted in ordering the wrong elements entirely.
These are the kinds of errors that cost time and money. If you order parts based on wrong numbers, you get wrong parts. If you design structural connections around a part that does not have the geometry you assumed, the connection fails when you try to build it physically. The rule now is absolute: every part number in the Stud.io file gets verified on BrickLink before the floor module is considered complete. The part's image, dimensions, and connection points must match what the design requires.
This sounds tedious. It is tedious. It has also saved me from ordering hundreds of dollars in wrong parts at least twice. The discipline of verification is not optional on a project at this scale.
Rule: never guess part numbers. #34146 was fabricated. #62113 was misidentified. Every part gets verified on BrickLink before the design is final. No exceptions.
Every detail described in this article exists because of a reference photograph, a video screenshot, or a cross-referenced measurement derived from multiple sources. The level of detail I am pursuing demands a reference library to match — 150-plus images, video sources, and careful cataloguing of what each source reveals about the building.
In Part 6: The Reference Hunt, I cover how that library was built from scratch — including the IMS Behind the Bricks video series that provided interior angles no still photograph could capture. When you have zero floor plans, your reference collection is everything.