Interior Computer Studio II

Before any software, the geometry itself. The taxonomy an interior designer actually meets: mesh (polygon) models — explicit vertices, edges and faces, always an approximation, a sphere is faceted no matter how fine; boundary-representation solids that carry a topological guarantee of closure, so the model KNOWS it encloses a volume; CSG solids that are a Boolean recipe rather than a surface; surfaces that are zero-thickness sheets with no enclosed volume; NURBS patches that are exact, resolution-independent analytic curvature (Rhino's world, and the right tool for curved joinery); subdivision surfaces with an editable cage; point clouds — unconnected measured points from LiDAR or photogrammetry, no surface at all, and increasingly relevant to Indian fit-out and retrofit where you model into an existing shell; and parametric/BIM objects that carry semantics. Then the honest centre of the unit: SOLID versus MESH. The real difference is a topological guarantee, not a shape — and it is inherited from manufacturing CAD, where it decides whether a part can be machined. For an interior designer it matters in exactly three places: CNC and joinery, 3D printing (which demands watertight geometry), and volume take-off. For the whole rest of the course — visualization — everything is mesh anyway, because renderers only ever trace triangles; the moment you hit Render, your NURBS and solids are tessellated, and that tessellation is lossy and irreversible. So the distinction that actually bites is not solid-versus-mesh; it is WATERTIGHT versus LEAKY and CORRECT versus FLIPPED NORMALS — and SketchUp geometry is notoriously non-manifold, which is why a SketchUp model so often refuses to 3D-print. The unit teaches the taxonomy so a student can read a file, and then tells the truth about which parts of it will ever matter to them.

A tour of the tools, framed as depth over breadth and stripped of marketing. What each is genuinely FOR: SketchUp — fast conceptual modelling, the default of the small Indian studio; Blender — a free, complete pipeline (model, sculpt, render, animate) whose licence is the real lesson; Rhino — precise NURBS for curved and product work; 3ds Max — high-end offline archviz; Revit — documentation and BIM, not visualization; and the renderers that plug into them — V-Ray, Corona, Enscape, Lumion, D5, Twinmotion, KeyShot. The cost-of-entry argument, told honestly and WITHOUT prices this course cannot verify: for a STUDENT the argument mostly dissolves — Autodesk is free for a renewable education year, Lumion Pro is free for students, Twinmotion is free under a revenue threshold, D5 has a free tier, and Blender is free forever. Cost is not a barrier to LEARNING. It bites the small Indian practice on graduation, which is much of why small studios sit on SketchUp. And Blender's GPL is categorically different: it is not a concession that expires when you graduate or cross a revenue line — it is a freedom that cannot be withdrawn, and the artwork you make is entirely yours (the GPL covers the application, not your output). That is a lesson about DEPENDENCY, not price. Then the myth that governs Indian syllabi: 'you must learn 3ds Max to be employable.' The evidence for it is circular — the surveys quoted are RENDERING-engine surveys that contain no modelling-software share at all, and the famous '3ds Max 59%' figure is an unmethodologised poll on a render-farm's blog, a population defined by paying to render heavy offline scenes. India's real job market is documentation-led: AutoCAD and Revit dwarf everything, and in interior-specific listings SketchUp is named about twice as often as 3ds Max. Learn 3ds Max if you want high-end studios; it is free while you are a student; it is not a gate.

How a model actually moves through a real project, and where it silently bleeds. The interop trap first: every exchange format loses something, and the danger is not that it loses data but that it loses it SILENTLY — no dialog says 'I dropped your cameras.' A format-by-format reckoning — OBJ (geometry only, no units, Phong-era materials), FBX (rich but proprietary and version-cursed), glTF/GLB (PBR-native, metres, +Y up — the modern delivery format), DWG/DXF (drawing exchange, poor 3D), SKP (keep it native), IFC (data not looks), STL (fabrication only), and dead 3DS. The right answer per hand-off: SketchUp to V-Ray or Enscape means don't export at all; modeller to web, game or AR means glTF; modeller to 3ds Max means FBX with units set FIRST; to fabrication means STL or DXF; to coordination means IFC. The units problem behind the classic 1/25.4: SketchUp works internally in inches and 3ds Max's default system unit is also an inch, so the 25.4 that appears as a mysterious scale factor is just inch-to-millimetre surfacing — set your system units before importing anything and never change them mid-scene, and check the RECEIVING end, not the sending one. Then the single most useful design rule in the course: the SILHOUETTE is the dividing line between what must be real geometry and what can be a normal map. Normal maps fake surface detail for light but do not move vertices, so silhouettes stay flat. For Indian interiors that means cornices, skirting, chair rails, jaali and lattice screens, carved teak and turned legs must be REAL geometry — you see them against the light — while plaster texture, fabric weave, wood grain, grout lines and stone pitting should be normal-mapped. And the myth that 'more polygons is always better,' which inverts at the top end: past pixel density a GPU shades tiny triangles up to eight times over for zero visible gain (Fatahalian et al., SIGGRAPH 2010), which is exactly why bake-high-to-low and normal mapping exist. PBR and why it made materials portable — roughness as a microfacet distribution, metallic as a near-binary switch, and glTF's metallic-roughness standard so a material means the same thing in two engines.

How a model becomes an image, and how to make that image honest. Rasterization projects triangles and asks 'which pixels?' — fast, light faked; ray tracing asks 'what does this pixel see?'; path tracing extends ray tracing to a Monte Carlo estimate of the full rendering equation, which is why noise exists — noise is estimator variance, and halving it costs four times the samples. Why real-time changed practice is the LOOP, not the pixel: offline is model, render, wait, look, fix; real-time is look while you model, and the design conversation moved into the client meeting. PBR materials — roughness is the statistical width of a microfacet distribution, not a bump map; metallic collapses two physical facts (metals have no diffuse and tinted specular; dielectrics have diffuse and ~4% untinted specular) into one slider; and PBR made materials PORTABLE, which is the real reason it won. Lighting with real photometry: an IES file (ANSI/IES LM-63) is an ASCII file of candela values measured on a goniophotometer by the luminaire manufacturer — the one place in your render where a laboratory speaks — and the awkward India finding that Wipro and Havells do not publish self-serve IES downloads, so the working route is the Philips/Signify India selector, global manufacturers like ERCO, or emailing the technical helpdesk, which is itself a normal professional act. The camera and the lie you must own: a wide lens does not add space, it compresses more solid angle into the frame and exaggerates depth, so a 3 × 3 m bedroom at 16 mm reads as generous — the honest interior range is about 28–35 mm and anything wider is a claim you are making. Exposure as real ISO, shutter and f-stop; two-point versus three-point and why parallel verticals read as a document while convergent verticals read as a snapshot. And the honest boundary of real-time engines: they take genuine IES input but the right word is UNVALIDATED, not inaccurate — plausibility is an input property, validation is an output property, and even Radiance publishes where it fails while a real-time tool's confident lux gradient has simply never been tested in public.

The project unit, where the model is built, lit and presented — and where the industry's favourite claim about 3D is tested against the evidence. First the workflow of a project: brief, model, material, light, camera, render, present, and the iteration that real-time makes cheap. Then the honesty reckoning the whole course has been building toward. The claim that 'moving to 3D reveals design flaws' is the weakest-supported idea in the field: the famous statistic — up to 40% fewer unbudgeted changes — traces, through a peer-reviewed paper, to a citation of a Stanford web-page INDEX, not a study, and appears in none of the reports it supposedly comes from; the two experiments that actually counted planted errors found NO significant difference between paper, monitor and VR, and what VR did change was speed, not thoroughness — reviewers were faster, not more accurate; and clash detection, routinely cited as proof, is automated geometric checking on hundred-crore hospitals with three subcontractors' ductwork colliding, which is MACHINE noticing, not a human noticing more in 3D, and has nothing to do with a student modelling one room. The soft, intuitive version — a designer notices a proportion problem in 3D they would have missed in plan — is folklore, and the honest thing is to say exactly that. What the evidence DOES support: the render as PERSUASION (the tradition runs unbroken from architectural shades-and-shadows manuals of 1904 to moving the sun in software today), realism as a treatment variable viewers cannot see, and abstraction as a leveller that narrows the expert/non-expert gap and invites discussion. The ethics of the persuasive image — a real published code of ethics for visualization exists — and spatial ability as a genuine, measurable predictor of interior-design performance. And the professional vacuum that frames the student's whole career: India has adopted no ISO 19650, BIS lists no matching standard, and there is no Indian standard for 3D or visualization deliverables at all — so your outputs are governed by nothing but client expectation and studio convention. That precise silence is the lesson: it is a professional space you will spend a career learning to navigate.