Photoreal rendering describes the goal of 3D graphics to produce an image that could plausibly be a photograph. It is less a single technique than an outcome defined by a set of physical accuracy criteria. These criteria include how light scatters across a surface, how shadows fall and soften with distance from their caster, how transparent materials refract the scene behind them, and how a material's finish responds differently to direct versus ambient light.
The term is widely used across industries that depend on visual accuracy for commercial or technical purposes: product visualization, automotive design, architecture, film visual effects, and e-commerce. In each context, photorealistic rendering serves the same purpose. It allows viewers to form accurate expectations about how something will look in reality before they have seen it in person.
Achieving photorealism requires that a rendering system simulate physical phenomena with enough fidelity that the human visual system, which is highly attuned to detecting anomalies in lighting and material behavior, cannot identify the output as computer-generated. This is a high bar. Human observers readily perceive incorrect shadow softness, missing ambient occlusion in crevices, implausible material reflectance, or color temperatures that do not match the stated lighting environment.
Light in the physical world follows principles described by rendering theory, particularly the rendering equation introduced by James Kajiya in 1986. The rendering equation expresses the total light leaving a surface point in a given direction as the sum of light emitted by the surface itself and light reflected from all other directions, weighted by the surface's reflectance properties. Solving this equation (or approximating it closely enough) is the computational challenge of photoreal rendering.
Ray tracing is the rendering technique most directly grounded in the physics of light. A ray tracer traces the path of light rays from the camera through each pixel into the scene, computing interactions with surfaces, spawning secondary rays for reflections and refractions, and sampling light sources to determine shadows. Path tracing extends this by stochastically sampling multiple light paths per pixel, averaging results to produce globally illuminated images in which indirect light (light that has bounced off one or more surfaces before reaching the camera) is realistically included.
Offline rendering systems, such as those used in film and high-quality product visualization, use path tracing with thousands of samples per pixel, running for minutes or hours to produce noise-free results. Real-time rendering, used in games, interactive product viewers, and web applications, has historically relied on rasterization (the process of converting 3D models, composed of vertices and polygons, or 2D vector graphics into a 2D grid of pixels) with approximations of global illumination effects. Hardware-accelerated ray tracing, available in modern GPUs following the introduction of NVIDIA's RTX architecture and standardized in Microsoft's DirectX Raytracing (DXR) API, has enabled physically-based real-time ray tracing in production.
Physically Based Rendering (PBR) is the material and shading framework that enables consistently and predictably photoreal results. Where earlier rendering approaches used artistic shading models that could look plausible in specific lighting conditions but broke down under others, PBR defines material properties in physically meaningful terms: how much light a surface reflects specularly versus diffusely, how rough or smooth the microsurface structure is, and whether the material is a metal or a dielectric. Materials defined using PBR parameters behave correctly across any lighting environment, a prerequisite for photoreal results in production contexts where lighting conditions are not fixed.
The glTF 2.0 format, the primary standard for delivering 3D content on the web, encodes material properties using the PBR metallic-roughness model defined by the Khronos Group. This standardization ensures that a material authored in a tool like Adobe Substance 3D renders consistently in any glTF-compatible viewer or engine.
Photoreal rendering quality has historically been confined to contexts where rendering time and hardware cost are not constraints — film production, high-end industrial visualization, and architectural rendering services. The rendering quality achievable in a web browser has been limited by the need for real-time performance on consumer hardware, using approximations of the physics that offline renderers compute exactly.
Two converging developments are changing this. First, physically based shading is now standard in real-time rendering, meaning web-based 3D experiences can use the same material definitions as offline renders and produce results that are physically consistent even if not identically detailed. Second, streaming architectures enable high-fidelity assets to be delivered progressively, so the quality ceiling for web-based 3D is no longer determined by download size. A photoreal product model, with high-resolution PBR textures, accurate geometry, and complex material layers, can stream to a mobile browser and render with physically based shading, producing an experience that closely approximates the quality of a dedicated product visualization tool.
For industries where visual accuracy directly influences purchasing decisions, return rates, or design sign-off, this convergence of photoreal quality with web-scale delivery represents a significant practical shift.
3D physically based rendering (PBR) — The material and shading framework that underpins physically accurate, photoreal rendering results.
Asset optimization — The process of preparing high-fidelity photoreal assets for efficient streaming and real-time delivery.
Gaussian splatting — A 3D representation method that captures photorealistic scene appearance from real-world photographs.
Photogrammetry — A capture technique that produces the high-fidelity geometry and texture data that photoreal rendering requires.
Streaming video reshaped media consumption; we're doing the same for 3D. Join a small team solving tough spatial streaming, content delivery, and developer experience challenges.
Technical deep-dives on adaptive spatial streaming, the infrastructure challenges of 3D at scale, and what we're learning as we build. Written by the team doing the work.
