Ray tracing is a class of rendering algorithms that simulates light by mathematically following individual rays as they travel through a 3D scene. The classical implementation starts at the camera, fires rays out into the scene, and computes what each ray hits — and at every intersection, secondary rays are spawned to figure out shadows, reflections, and refractions.
Ray tracing is the foundation of photorealistic rendering. It produces accurate reflections, refractions, shadows, and (when combined with bounce sampling) global illumination. It's also computationally expensive — a single ray-traced architectural image at high quality can take minutes to hours on a workstation, which is why traditional rendering tools like V-Ray and Lumion lean on optimization tricks and GPU acceleration.
AI rendering takes a different approach: instead of computing ray paths from physics, the AI model has learned what ray-traced images look like from a training set, and produces statistically similar results in seconds. The output isn't 'real' ray tracing — no rays are traced — but the result is visually equivalent for most practical purposes. The trade-off is unpredictability: AI renders can occasionally produce light behavior that wouldn't survive a physics check, while a true ray-traced render is always physically correct.
See also