When talking of autostereoscopy types, we need first need to refer to the type of display that generates the effect: Two-view displays and multi-view displays.
- Two-view displays: these are the ones that only send information about two images to our brain. Whether using parallax barrier or lenticular lenses. In this type of displays, each eye receives one specific image (left image to left eye, for example)
- Multi-view displays: These are somewhat special, the viewer only receives the information about two images, but has the possibility to move horizontally and watch a new stereo pair (two new images that will be processed as 3D by the brain, or the same scene from different perspective, chose the one easier to understand). This results in the possibility to move horizontally and watch different perspectives of the same object resulting in obtaining more information about the object and thus more understanding. The number of perspectives depends on the number of views the display has. Also, these displays allow multiple people to watch the 3D scene simultaneously.
Autostereoscopy can be segregated into types according to the physical technique used to create the effect. Two main techniques are used: parallax barrier and lenticular lenses. Each of the types has a way of encoding the information of the n views so that the light paths are modified in such a manner that one eye receives different information than the other eye. This encoding is called interleaving. It consists of creating one image that stores information about the pixels in each of the n views with the less information loss possible and in some sort of arrangement. That arrangement is called the interlacing/interleaving pattern that is consequent with the physical method used for the 3D view.
Let’s clarify this.
- Parallax barrier based autostereoscopy: this kind of technique uses light blocking to avoid one eye to receive information of one of the images. The technique consists of an opaque screen that has slits that are transparent. This way, each eye will receive only one image at a time, since columns of the display pixel matrix are blocked. The image encoding is quite simple: combine both images into one alternating one column of each.
the new image will be wider (twice exactly) than the original. In a two view autostereoscopic this technique requires the viewer to locate at a sweet spot (a certain distance at which the viewer must be in order for the information to arrive in each eye). Some times, this technique is used along a head-tracking display to allow the movement of the head.
Multiview parallax barrier based autostereoscopic displays exist, however in order to see different stereo pairs, one must remain at a fixed position.
- Lenticular lenses: this technique uses refraction to send each eye information of only one image. This displays consist of an LCD panel that has a curved lenslet array in front of it. Depending on the angle of viewing the eye will receive information about one LCD pixel or another. Lenticular lenses can be installed vertically or in a slanted way (which reduces the loss of information). Depending on the type of lens installation, the interlacing pattern to combine the n views will change. This type of autostereoscopy allows the horizontal movement of the viewer and multiple and simultaneous viewers.