Photon mapping is similar to the ray tracing process. While ray tracing is based on rays from the observers/camera position, photon mapping is based on rays emitted from the light source. Photon mapping uses virtual "photons" radiating light into the room. When they hit a surface, they are reflected back and the luminance values are summated. The photon outputs are stored in a photon map. This map is not bound to the geometry and can be used for simulations with distributed calculations in the network. The camera position can be modified without the need to revise the calculation - this process, though, is not interactive.
The more photons a model has, the more accurate the transitions will be in the rendering and the more complex the calculation. After a certain number of reflections/iterations, the photon map has the required precision. In a further process, the points can be merged through gathering.
Photon mapping is used as for further calculations. To show details more accurately, the process is combined with ray tracing. If the calculation is based exclusively on ray tracing it is too complex for very small models and very bright light sources.