Ray tracing is a rendering technique that synthesizes an image by tracing the paths of light rays through a three dimensional scene. In its most common form rays are cast from the virtual camera through each pixel of the image plane into the scene; where a ray strikes a surface, the renderer evaluates how much light arrives there, often by casting further rays toward light sources and in reflected or refracted directions. Because it works in terms of straight light paths, ray tracing handles mirror reflection, transparency, and shadows within a single coherent model rather than as separate effects bolted onto a local shading calculation.
The idea of casting rays to determine what is visible and how it is lit goes back to Arthur Appel’s 1968 paper “Some Techniques for Shading Machine Renderings of Solids,” presented at the AFIPS Spring Joint Computer Conference, which first used rays to resolve visibility and to test for shadows. The technique became a complete rendering method with Turner Whitted’s 1980 paper “An Improved Illumination Model for Shaded Display,” which made the process recursive: each ray that hits a reflective or refractive surface spawns secondary rays, building a tree of light paths per pixel. This recursive scheme, now called Whitted or classical ray tracing, produced the first images with accurate interreflection between glass and mirrored objects.
For decades ray tracing was an offline method. It is computationally expensive because each pixel may require tracing many rays, each of which must be intersected against the scene geometry, and high quality images need many samples per pixel to reduce noise. Film and product visualization adopted it for its physical accuracy, while interactive applications such as games relied instead on rasterization, which projects geometry directly to the screen and is far cheaper per frame. Acceleration structures such as bounding volume hierarchies and spatial trees reduced the cost of finding ray and surface intersections, but the method remained too slow for real-time use on general hardware.
Ray tracing also underpins more physically complete rendering. The reflection, refraction, and shadow rays of Whitted’s model capture specular transport but not the diffuse indirect light that bounces between surfaces. Path tracing, derived from the rendering equation, extends ray tracing into a full solution of global illumination by randomly sampling many light paths and averaging the results, at the cost of additional noise that must be filtered or denoised away.
Real-time ray tracing arrived on consumer hardware in 2018, when GPUs began shipping with dedicated ray and triangle intersection units and graphics APIs exposed ray tracing pipelines to developers. This let interactive applications combine rasterization for primary visibility with ray traced reflections, shadows, and global illumination, blurring the long standing divide between offline and real-time rendering. The canonical primary sources remain Appel’s 1968 paper and Whitted’s 1980 paper, which together define the technique.