Why Materials and Lighting Decide the Result

Geometry gets you a plausible shape, but it is the surface response and the light falling on it that convince a client, a planning officer or a fabrication team. A common mistake is to spend days modelling a bracket or a facade and then accept the default grey material. Two settings — roughness and reflectance — plus the physical size of your light sources will change more about the final image than another fifty thousand polygons ever will. The good news is that this is not guesswork. Real materials have measurable behaviour, and most renderers in use across UK practice give you direct control over it.

Roughness: The Setting That Does Most of the Work

Roughness describes how much a surface scatters light. It is not a synonym for "glossy" alone; it governs the width of the highlight and the sharpness of reflections. Work in real terms:

• 0.05–0.15 — polished glass, lacquered brass, chrome, polished stone.
• 0.25–0.4 — satin paint, anodised aluminium, sealed timber, powder-coated steel.
• 0.5–0.7 — brushed metal, matt emulsion, concrete, brick.
• 0.8–1.0 — chalk, plaster, felt, unfinished MDF, carpet.

Uniform values are the giveaway. A single roughness number across a whole wall produces a plastic sheen. Add variation with a grunge or noise map, held to a narrow range — say 0.4 to 0.55 for render, 0.6 to 0.8 for brick — and the surface starts reading as a real material with patches and history.

Reflectance, Metalness and the Fresnel Effect

Reflectance is the strength of the specular response. In most physically based workflows you control it through an index of refraction or a specular level. Sensible starting points: 1.5 IOR for most plastics and painted surfaces, 1.33 for water, 1.52 for window glass, and higher values for ceramics and gems. Change it only when you have a reason.

Metalness should be treated as a switch. It is 1 for bare metal and 0 for everything else. Painted steel, galvanised sheet with a topcoat, and dusty aluminium are all non-metals — the metal is buried under the coating, and the coating is what you see. Values like 0.5 are almost never correct and tend to produce a muddy, unplaceable surface. Remember too that metals have no diffuse colour: their base colour tints the reflection itself. Copper, brass and gold will look wrong if you leave the base colour grey.

Fresnel behaviour matters at grazing angles, which is why a floor reads brighter when you look along it and why glazing appears more reflective than a plan view suggests. This happens automatically in a physical shader, but you will flatten it if you push reflectance up to compensate for a badly lit scene.

Lighting Balance: Key, Fill and Temperature

Lighting a visualisation is closer to studio photography than to drawing. Start with a key light, add a fill at roughly a quarter to an eighth of the key's intensity, and use a rim or a practical source to separate the subject from the background. Keep model units in metres so that inverse-square falloff behaves properly; if your scene is in millimetres, every light will look wrong.

Colour temperature does a lot of quiet work. Daylight sits around 5600 K, overcast around 6500–7000 K with a cool cast, and tungsten practicals near 3200 K. Mixing them is fine and often desirable — a warm interior against a cool window view reads immediately as a real building. Check your camera exposure rather than cranking light intensity, and watch the histogram for clipped highlights.

Shadow Softness and Contact

Shadow softness comes from the size of the light relative to the subject, not from a slider on its own. A 1.2 m softbox two metres from a chair gives a soft edge; a point light gives a hard one. If you are rendering a window into a room, model the opening as a large area light or use a portal so the shadow pattern has the right shape and sharpness. Shutter slats, structural ribs and deep reveals only read properly when the light source matches the opening.

Watch for the two common failures. Floating objects, where a chair or kerb appears detached because the contact shadow is missing — a low ambient occlusion pass or a dedicated contact shadow will settle it. And peter-panning, where the shadow lifts away at the base of a wall because bias is set too high; reduce it until the artefact disappears rather than leaving a safety margin in place.

Building a Repeatable Check Routine

Set up a routine and it becomes fast:

• Render at low samples and small resolution while you tune materials and lighting; only push quality at the end.
• Keep a lighting rig file with your standard key, fill and environment, then merge it into each new project.
• Photograph a real sample of the material in overcast light and compare its luminance with your render on screen.
• Save material presets as you refine them — a library of ten well-judged surfaces beats a hundred generic ones.
• Check the result on a second display or a print before you send it out, because a bright monitor flatters soft lighting.

None of this requires exotic tools. It requires patience with three or four settings, a metre stick, and a willingness to compare your render with the thing it is meant to represent.