Raytracing with Backlight

Backlight web site

This page describes how the movies on this site were produced.

The movies are merely sequences of images, so we shall just describe how an image is produced. The method is called "relativistic raytracing", and is implemented in software called "Backlight" written by Antony Searle. The underlying concept is to use fundamental optical and relativistic physics to follow the behaviour of the light which forms the images.

Images are taken by a notional camera, often refered to as "your eye" on this site. The camera is modelled as a pinhole camera, with a small "lens" and an array of "pixels" as the film. This is the usual setup for raytracing, a technique commonly used to produce realistic computer generated images, including in popular films.

The image is actually generated backwards: a fictional light pulse, or "photon" is assumed to leave each camera pixel at the time the image is taken. This photon travels in a straight line out of the pinhole lens into the scene. The scene contains objects which may be moving relative to the camera. The photon is transformed into the rest frame of each object using the equations pertaining to relativistic aberration. This leaves us with a photon travelling in a known direction relative to an object at rest. It is straightforward to determine whether or not the photon will strike the object.

If it does, then we can deduce that the photon originally left from that object. Light from the intersection point must have reached the camera pixel, along the reversed path of the fictional photon coming from the camera. Standard raytracing techniques, described below, are then used to find the colour and intensity of this light based on the colour and reflective properties of the object. Finally, the true colour and intensity of this light is found by transforming back into the camera rest frame using the relativistic doppler and intensity transformations.

This process is repeated for each object's rest frame, and for each camera pixel, to build up the complete image. In the desert road example below there are only two frames: the camera frame and the road frame. The raytracing is done in the road frame, and then the pixel light is transformed to the camera frame.

Raytracing is a highly developed technique for computer generating realistic images. An excellent freeware raytracer is POVRAY. The basic idea is to follow light rays from light sources, via the objects in the scene, to the camera. At each object's surface the behaviour of the light ray may be specified to mimic particular behaviours or textures, such as mirrors or paint. Light may reach an object directly from light sources, or after interacting with other objects. This allows for shadows and for illumination by scattered light.

Althought raytracing can produce relaistic images, it is computationally intensive. The more interactions light is allowed to have before reaching the camera, the more realistic the image, but the more computation is required.