Inferensys

Glossary

Physically Based Rendering (PBR)

Physically Based Rendering (PBR) is a shading and rendering methodology that models the interaction of light with surfaces using realistic physical properties, such as albedo, metallicness, and roughness, to achieve high-fidelity visual results.
Stylish WeWork-like workspace with hot desks and document wall, professional searching through enterprise knowledge base on a mounted ultrawide display, warm industrial pendants overhead.
SIMULATION ENVIRONMENT GENERATION

What is Physically Based Rendering (PBR)?

Physically Based Rendering (PBR) is a shading and rendering methodology that models the interaction of light with surfaces using realistic physical properties, such as albedo, metallicness, and roughness, to achieve high-fidelity visual results.

Physically Based Rendering (PBR) is a computer graphics methodology for shading and lighting that uses physically accurate models of light transport and surface reflectance. Unlike older, artistic approaches, PBR relies on measured material properties—primarily albedo (base color), metallicness, and roughness—to determine how a surface interacts with light. This ensures visual consistency across different lighting environments, a critical requirement for creating believable simulation environments used in robotics training and digital twins.

The core of a PBR workflow involves a bidirectional reflectance distribution function (BRDF), a mathematical model that describes how light is reflected at a surface point. Common models include the Cook-Torrance microfacet BRDF. By adhering to physical laws like energy conservation and Helmholtz reciprocity, PBR produces predictable, realistic results. This technical rigor makes it the industry standard for real-time engines and offline renderers, enabling the creation of high-fidelity virtual worlds for sim-to-real transfer learning and synthetic data generation.

SIMULATION ENVIRONMENT GENERATION

Core PBR Material Properties

Physically Based Rendering (PBR) uses a standardized set of material properties to model light interaction with surfaces, enabling high-fidelity visuals essential for creating realistic simulation environments.

01

Albedo (Base Color)

The Albedo map defines the intrinsic color of a material, representing the diffuse reflectance of a surface under pure white light. For dielectric (non-metallic) materials like plastic or wood, this is the base color. For pure metals, the albedo should be black or near-black, as metals absorb all diffuse light; their color comes from specular reflection. This map should be free of lighting information (shadows, ambient occlusion) to maintain physical accuracy.

  • Key Rule: No baked lighting or shadows.
  • Example Values: Concrete (mid-gray), grass (green), copper (black for albedo, with color from specular).
02

Metallic

The Metallic map is a grayscale mask that defines whether a surface is a metal (white) or a dielectric/non-metal (black). This binary or near-binary property is crucial as it controls how the renderer interprets the albedo and reflectance values. Metals conduct electricity, meaning light does not penetrate; their albedo defines the color of specular reflections. Dielectrics are insulators, with albedo defining diffuse color and a constant low specular reflectance.

  • Workflow: Uses a Metallic/Roughness workflow.
  • Typical Values: 1.0 (pure metal), 0.0 (non-metal), values between 0-1 for corroded or painted metals.
03

Roughness

The Roughness map controls the micro-surface detail of a material, determining how light scatters upon reflection. A low roughness value (0.0, black) creates a mirror-like, sharp specular highlight. A high roughness value (1.0, white) creates a broad, matte highlight that blurs reflections. This property is perceptually inverse to Glossiness; a roughness of 0.3 is equivalent to a glossiness of 0.7.

  • Physically Based: Models microscopic surface imperfections.
  • Visual Effect: Directly affects the sharpness of reflections and the size of specular highlights.
04

Normal Map

A Normal Map is an RGB texture that simulates high-resolution surface detail (like bumps, grooves, or scratches) without adding geometric complexity. Each pixel's color (R,G,B) corresponds to a surface normal vector (X,Y,Z), which perturbs how light interacts with the low-poly mesh. It creates the illusion of depth and detail. A Height Map (displacement map) is a related grayscale texture that can actually displace geometry, but is more computationally expensive.

  • Purpose: Fake high-frequency surface detail.
  • Color Meaning: RGB = (X, Y, Z) vector in tangent space.
05

Ambient Occlusion (AO)

An Ambient Occlusion map is a grayscale texture that simulates soft shadowing in crevices and areas where ambient light is occluded. It darkens these contact points, adding crucial depth and grounding to objects. While AO can be calculated dynamically, a baked AO map is a performance-efficient way to add this contact shadowing for static elements. In a PBR workflow, the AO map is typically multiplied with the diffuse/albedo contribution.

  • Effect: Adds contact shadows and depth.
  • Usage: Multiplied with the final diffuse lighting.
06

Emissive

The Emissive map defines parts of a material that emit their own light, independent of scene lighting. This is used for materials like screens, LEDs, or neon signs. The map's RGB values define both the color and intensity of the emitted light. In modern engines, emissive materials can act as dynamic light sources, contributing to scene illumination through techniques like Global Illumination, making them vital for creating realistic light fixtures within simulations.

  • Use Case: Light-emitting surfaces.
  • Advanced Effect: Can contribute to real-time GI (Screen Space GI, Lumen, etc.).
SIMULATION ENVIRONMENT GENERATION

PBR Workflows: Metallic/Roughness vs. Specular/Glossiness

Within Physically Based Rendering (PBR), two primary material workflows define how surface properties are encoded in texture maps, each with distinct advantages for asset creation and rendering performance.

The Metallic/Roughness workflow uses a simplified, artist-friendly material model defined by three core maps: a base color albedo map (RGB), a grayscale metallic map (where white is pure metal), and a grayscale roughness map (where white is fully rough). This model assumes non-metallic surfaces have no specular color, deriving reflectivity from a constant Fresnel value, which streamlines authoring and is the dominant standard in real-time engines like Unity and Unreal.

The Specular/Glossiness workflow offers greater explicit control by separating the specular reflection color from the diffuse albedo, using a specular map (RGB), a diffuse albedo map (RGB), and a grayscale glossiness map (where white is perfectly smooth). This model can more accurately represent certain non-metallic materials with colored reflections, like plastics or gems, but requires more careful authoring to maintain physical plausibility and is less common in modern real-time pipelines.

COMPARISON

PBR vs. Legacy Shading Models

A technical comparison of Physically Based Rendering (PBR) and traditional, non-physical shading models, highlighting the fundamental differences in approach, inputs, and visual outcomes.

Feature / PropertyPhysically Based Rendering (PBR)Legacy / Ad-Hoc Shading (e.g., Phong, Blinn-Phong)

Underlying Principle

Models light-surface interaction using energy conservation and microfacet theory derived from physical optics.

Uses empirical, artist-driven models that approximate visual phenomena without strict physical basis.

Core Material Inputs

Albedo (base color), Metallic, Roughness/Smoothness. These are (mostly) invariant under different lighting.

Diffuse Color, Specular Color/Level, Glossiness/Shininess. These are often lighting-dependent and require tuning per scene.

Energy Conservation

Fresnel Effect

Accurately modeled. Reflectivity increases at grazing angles for all materials.

Often approximated or omitted. Requires manual tweaking of specular falloff.

Lighting Response

Materials respond predictably and consistently to any HDR environment map or light setup.

Materials require manual re-tuning when lighting conditions change to look 'correct'.

Authoring & Workflow

Asset creation is more constrained and measurable. Values can often be sourced from real-world measurements.

Asset creation is highly artistic and subjective. Relies heavily on the artist's eye and scene-specific adjustments.

Artistic Control

Control is indirect via physical parameters. Achieving specific non-physical 'stylized' looks requires breaking the model.

Direct, unbounded control over visual appearance. Easier to achieve exaggerated, cartoonish, or fantastical looks.

Portability Across Engines

High. A PBR material authored in one engine (e.g., Unreal) will look largely correct in another (e.g., Unity).

Low. Materials require significant re-authoring and tweaking when moved between different engines or renderers.

Performance Cost (Modern GPUs)

Comparable or optimized. Modern engines and hardware are designed for PBR pipelines.

Can be lower for very simple models, but often comparable when achieving similar visual quality.

Primary Use Case

Photorealism, consistency, and predictable asset pipelines for simulation, film, and AAA games.

Stylized graphics, legacy game engines, or scenes where full physical accuracy is not the goal.

PHYSICALLY BASED RENDERING (PBR)

Frequently Asked Questions

Physically Based Rendering (PBR) is a shading and rendering methodology that models the interaction of light with surfaces using realistic physical properties, such as albedo, metallicness, and roughness, to achieve high-fidelity visual results. These FAQs address its core principles, implementation, and role in simulation.

Physically Based Rendering (PBR) is a computer graphics methodology that simulates the flow of light by modeling material properties according to physical laws, rather than using artistic approximations. It works by using a set of standardized, measurable surface attributes—primarily albedo (base color), metallicness, and roughness—as inputs to bidirectional reflectance distribution functions (BRDFs). These BRDFs, such as the Cook-Torrance or GGX models, mathematically describe how a surface reflects light based on its microgeometry and composition. The system calculates lighting by conserving energy (light reflected cannot be brighter than the light received) and ensuring physical reciprocity. This produces consistent, predictable visuals under any lighting condition, which is critical for creating believable simulation environments for training AI and robotic systems.

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.