Roughness Values — The Setting Most 3D Artists Get Wrong | Sameer Baloch
Look Development
12 min read

Roughness Values
The Setting Most 3D Artists Get Wrong

Roughness is the most important material setting in any PBR renderer. It determines how light scatters across every surface in your scene. Most artists set it by feel, by memory, or by copying a value from a tutorial. All three approaches produce the same result: surfaces that look like plastic, rubber, or nothing specific at all.

Sameer Baloch
Senior 3D Environment and Lighting Artist
Maya / Redshift / UE5 / Blender / Cinema 4D
Software Covered

This article covers Maya with Redshift, Unreal Engine 5, Blender (Cycles and EEVEE), and Cinema 4D with Redshift. The roughness value ranges and the approach to reading reference apply identically across all PBR renderers. Where software-specific input names differ, each is called out.

01

What Roughness Actually Controls

Roughness controls how light scatters when it hits a surface. At a value of zero, the surface is a perfect mirror. Light bounces off at a precise angle and you see a sharp, perfect reflection of everything around it. At a value of one, the surface scatters light in every direction equally, like chalk or dry concrete. There are no reflections visible at all, only diffuse scatter.

Every real-world material exists somewhere on that scale. Glass is near zero. Brushed steel is around 0.3 to 0.45. Painted wall is 0.6 to 0.75. Dry sandstone is 0.85 to 0.95. The position of each material on this scale is not arbitrary or aesthetic. It is a measurable physical property of how that material interacts with light at the microscopic level.

In a PBR renderer, roughness is the primary variable that makes surfaces look like different materials. Two objects with identical colour, identical geometry, and identical lighting will look completely different from each other if their roughness values are different. It has more visual impact on material believability than any other single setting.

"Two surfaces with the same colour but different roughness values read as completely different materials. Roughness is identity."

02

Why 0.5 Is Almost Always Wrong

The default roughness value in most 3D software is 0.5. It is the starting point, the midpoint, the value that ships when you create a new material. And it is the least accurate value for almost every real-world material that exists.

A roughness of 0.5 produces a surface that is too shiny to read as a matte material and too rough to read as a reflective one. It sits in a visual no-man's-land that corresponds to almost nothing in the real world. It looks like slightly dirty plastic. Nothing more specific than that.

When an entire scene is built with materials sitting near 0.5, every surface starts to look the same. Stone and concrete look the same. Metal and painted wood look the same. The scene loses material differentiation entirely and reads as a collection of plastic objects under studio lighting.

The first thing to check when a scene looks like a plastic model: select five random surfaces and check their roughness values. If they are all between 0.4 and 0.6, that is your problem. Push the extremes. Rough things rougher. Smooth things smoother. Material differentiation is roughness differentiation.

03

How to Read Reference for Roughness

The only reliable method for setting roughness accurately is from real-world photographic reference. Not from memory. Not from approximation. From an actual photograph of the specific material under similar lighting conditions, examined closely for how light behaves on its surface.

When you look at a reference photograph to determine roughness, you are asking one specific question: how sharp are the highlights and reflections on this surface? The sharpness of the highlight is a direct visual representation of the roughness value.

What You See in ReferenceRoughness RangeMaterial Examples
Perfect mirror reflection, sharp edges0.0 to 0.08Mirror glass, polished chrome, still water
Near-perfect reflection, very slight blur0.08 to 0.18Polished metal, piano black, lacquered surface
Visible reflection, soft edges0.18 to 0.35Brushed metal, satin finish, wet stone
Broad highlight, faint reflection0.35 to 0.55Painted surface, leather, slightly worn plastic
Wide soft highlight, no visible reflection0.55 to 0.72Rubber, skin, matte painted wood
Very broad highlight, barely visible0.72 to 0.88Dry concrete, fabric, unfinished wood
No highlight visible, pure diffuse scatter0.88 to 1.0Chalk, dry sandstone, raw cotton
The Reference Method

Open a photograph of the material in Photoshop. Look at where the light source reflects on the surface. If you can see the shape of the light source clearly, the roughness is low, below 0.3. If the light source is visible but blurred into a large soft area, roughness is mid-range, 0.3 to 0.6. If there is no visible reflection and only a broad soft highlight exists, roughness is above 0.65.

04

Reference Values by Material Category

These are production-tested starting points. Always adjust from reference specific to your scene. Lighting conditions, surface age, weathering, and cleanliness all affect the actual value needed.

Metals

Polished chrome
0.02 to 0.08
Near-perfect mirror. Light source shape visible with minimal blur. Use as a specular-dominant surface.
Stainless steel, clean
0.1 to 0.2
Slightly blurred reflections. Light source still visible. Fine directional scratches may shift this higher.
Brushed aluminium
0.3 to 0.5
Directional blur from brushing. Reflections visible but stretched and soft in the brush direction.
Painted metal, gloss
0.08 to 0.18
Metallic 0 here. The paint surface determines roughness, not the metal beneath. Treat as a painted surface.
Rusted metal
0.65 to 0.82
Oxidation destroys reflectivity. Rust reads more like stone than metal at the micro level.
Raw cast iron
0.55 to 0.72
Grainy surface texture scatters light broadly. Visible highlight but no discernible reflection.

Stone and Concrete

Polished marble
0.05 to 0.15
Highly reflective when polished. Treat similarly to glass for roughness. Metallic stays at 0.
Honed stone
0.3 to 0.45
Matte but with visible sheen. Broad reflections with soft edges. Common for interior architectural surfaces.
Dry concrete
0.78 to 0.92
Almost no visible reflections. Very broad soft highlight only. Most beginners set this too low at 0.5 to 0.6.
Wet concrete
0.25 to 0.4
Water film on surface creates a near-mirror layer. Dramatically lower than dry concrete.
Rough sandstone
0.85 to 0.96
Pure diffuse scatter. No highlight visible at all in most lighting conditions.
Smooth river rock
0.35 to 0.55
Water erosion creates a relatively smooth surface. Broader reflections than rough stone but no sharp mirror quality.

Organic and Natural

Human skin
0.45 to 0.65
Subsurface scattering is the primary complexity here. Roughness around 0.5 to 0.6 for most skin types, lower for oily or wet skin.
Ice, clear
0.02 to 0.12
Highly transparent and reflective. Treat similarly to glass. Roughness varies dramatically with surface texture.
Snow, packed
0.82 to 0.94
Despite being made of ice crystals, snow scatters light in all directions. Very high roughness, near pure diffuse.
Dry wood, unfinished
0.75 to 0.88
Grain and fibres scatter light broadly. Barely any visible highlight. Often set too low by beginners.
Lacquered wood
0.05 to 0.15
The lacquer surface determines roughness entirely, not the wood. Similar to polished plastic.
Dry soil / dirt
0.88 to 0.98
Completely diffuse. No specular contribution visible. Push to near 1.0 for dry earth in direct sun.

Manufactured Surfaces

Glass, clear
0.0 to 0.05
Nearly perfect mirror on the surface. Transmission handles the transparency. Roughness controls the surface quality.
Frosted glass
0.3 to 0.55
Etched surface creates broad diffuse reflections. Transmittance also becomes scattered with frosting.
Gloss paint
0.05 to 0.15
Very smooth dried paint surface. Reflections visible and relatively sharp. Metallic stays at 0.
Satin paint
0.25 to 0.4
Soft sheen. Reflections visible but significantly blurred. Common interior wall finish.
Matte paint
0.6 to 0.78
Broad, barely visible highlight. No discernible reflection. Most common wall paint in interior environments.
Rubber, smooth
0.45 to 0.6
Slightly glossy. Broad soft highlight. Lower for new rubber, higher for aged or weathered rubber.
Fabric, cotton
0.85 to 0.96
Fibrous surface scatters almost completely. Near-pure diffuse. Fine fabrics like silk are dramatically lower.
Silk
0.15 to 0.35
Smooth filaments create a sheen. Directional highlights from weave structure. Very different from cotton.
Leather, new
0.3 to 0.5
Soft sheen from surface coating. Broad highlight. Aged leather with cracking moves toward 0.6 to 0.75.
Plastic, gloss
0.05 to 0.15
Very smooth surface. Sharp reflections. Metallic at 0. The surface most beginners accidentally reproduce with default values.
Plastic, matte
0.55 to 0.72
Broad diffuse highlight. No sharp reflections. Common for product surfaces, electronic housings.
05

The Metallic Rule

The metallic value in PBR is binary. It is either 0 or 1. A surface is either a metal or it is not. There is no physically accurate value between these two extremes for a pure material. Partial metallic values produce results that correspond to no real-world material and cannot be corrected by adjusting other parameters.

The reason is how the PBR model calculates light interaction. Metallic surfaces have no diffuse component. All light either reflects off the surface or is absorbed. Non-metallic surfaces have a diffuse component and very limited specular response. Values between 0 and 1 produce a hybrid that blends these incompatible lighting models in a way that does not exist in physical reality.

Metallic: 0 or 1. Never 0.3, never 0.5, never 0.7. The only exception is a transition blending map used at the boundary between a metal and a non-metal surface, such as rust covering steel. In that case, the metallic texture handles the transition, not a single partial value.

Common Metallic Mistakes

Painted metal set to Metallic 1. Once metal is painted, the surface is paint, not metal. The metallic value should be 0. The roughness should reflect the paint finish, not the metal beneath it.

Metallic 0.5 on any surface. There is no physical equivalent. It produces a result that looks like neither metal nor non-metal. If a surface looks wrong and the metallic value is between 0 and 1, set it to either 0 or 1 first and reassess.

Using metallic instead of roughness for shiny surfaces. Artists sometimes increase metallic value to make a non-metal surface look shinier. The correct approach is to lower the roughness value. A shiny piece of plastic has Metallic 0 and Roughness 0.05 to 0.15. Not Metallic 0.4.

06

The Six Most Common Roughness Mistakes

01 / Everything is set near 0.5
The entire scene uses default or near-default roughness values. Every surface reads as slightly dirty plastic. Stone looks the same as wood. Metal looks the same as concrete. There is no material differentiation and no visual interest from surface variation.
Fix
Select each major surface type and push them toward their actual physical roughness. Concrete up to 0.8 or higher. Polished metal down to 0.1. Fabric up to 0.9. Create contrast between surface types. The eye reads material variety through roughness contrast.
02 / Stone and rock set too low
Dry stone, concrete, and earth set to roughness 0.4 to 0.6. These surfaces have visible specular highlights and faint reflections in the render. In real life they have essentially none. The scene reads as wet or plastic rather than dry and physical.
Fix
Dry stone roughness starts at 0.75 and goes up toward 0.95 for very rough surfaces. Only add water or moisture to bring roughness down. A dry rock face in the sun has no visible reflections at all.
03 / Metal roughness set too high
Metal surfaces set to roughness 0.5 or higher. The render shows broad, flat highlights with no discernible reflection. The surface reads as grey rubber or painted plastic rather than any kind of metal. The metallic value is correct at 1 but the roughness is destroying the material read.
Fix
Polished metal starts at roughness 0.05 to 0.15. Even heavily worn and scratched metal rarely exceeds 0.5. Use roughness texture maps to add variation rather than raising the base value. Metal should show reflections.
04 / No roughness variation within a material
A single uniform roughness value across an entire object. No texture map, no variation, no weathering or wear. The surface looks synthetic and computer-generated because real surfaces always have microscopic variation in roughness across their area.
Fix
Use roughness texture maps even for simple materials. A concrete surface has roughness variation from 0.7 to 0.9 across its area. A metal surface has variation from 0.1 to 0.35 from scratches and wear. This variation is what makes materials read as real.
05 / Not accounting for surface condition
Using the roughness value for a clean, new material on a worn, aged, or dirty version of the same material. A clean car and a dusty car have dramatically different roughness values. A new leather boot and a worn one are not the same material in PBR terms.
Fix
Always reference the specific condition of the material, not just the material type. Age, dirt, moisture, wear, and coating all shift the roughness value significantly. A dusty surface gains 0.2 to 0.35 roughness. A wet surface loses 0.3 to 0.5 roughness.
06 / Setting roughness the same as similar-looking materials
Assuming that materials which look similar in colour should have similar roughness. Grey concrete and grey metal look similar in colour but have completely different roughness values. This confusion between colour similarity and material similarity is one of the most common causes of scenes where everything looks the same.
Fix
Roughness is determined by surface physics, not colour. Look at how light behaves on the surface in reference photographs, not at what colour the surface is. Grey concrete is 0.8. Grey brushed metal is 0.35. Same colour, completely different materials.
07

Where to Find the Roughness Input

Maya / Redshift

In the Redshift Standard Material, the roughness input is labelled Refl. Roughness under the Reflection section. This controls the width of specular highlights and the sharpness of reflections. A separate Coat Roughness controls an optional clear coat layer for materials like car paint.

Unreal Engine 5

In the Material Editor, the Roughness pin on the material node accepts a value between 0 and 1 or a texture map. The M_Basic master material workflow uses a texture parameter named Roughness that drives this input. Scalar Parameter nodes allow per-material instance control.

Blender / Cycles

In the Principled BSDF shader node, the Roughness input is the third parameter from the top of the main settings. It accepts values 0 to 1 or a connected texture node. For EEVEE, the same node and values apply. GGX roughness in Blender maps directly to roughness in other PBR renderers.

Cinema 4D / Redshift

In the Redshift Standard Material for C4D, the roughness input is in the same location as Maya Redshift since it is the same renderer. In the standard C4D material, roughness is controlled through the Reflection layer width settings, which is less intuitive but produces equivalent results.

The Input Name Does Not Change the Value

Whether the input is labelled Roughness, Refl. Roughness, Gloss, or Specular Roughness, the value 0.3 means the same thing in every PBR renderer. A roughness of 0.3 on a metal surface in Maya Redshift will look the same as roughness 0.3 on the same metal in Blender Cycles or UE5. The scale is universal.

The One Practice That Fixes Everything

Open a reference photograph of every major material in your scene before setting any roughness value. Look at where the light hits the surface. Ask: can I see the shape of the light source? Is it sharp, blurred, or completely gone? Set your roughness to match what you see. Do this once for every material in every scene and the plastic look will disappear permanently.

Roughness accuracy is not about memorising a table of values. It is about developing the habit of looking at reference and translating what you see into a number. That habit, applied consistently, is what separates materials that look like materials from materials that look like render defaults.

THE COMPLETE LOOK DEVELOPMENT WORKFLOW

The full workflow guide covers look development in depth: how roughness, reflectivity, and lighting work together as a system, with specific settings for Maya Redshift and UE5 side by side. Free to read online, or $29 for full access.

Read Free Guide Full Access, $29
Let's Work Together

GET IN TOUCH

Available for freelance projects, collaborations, and commissions. Based in Karachi, working globally.