Laboratories around the world have been investigating the 3D perception and human factors requirements necessary for high quality stereoscopic pictures. The most significant human factors studies to date have been performed at NHK (Japan), CRC (Canada), HHI (Germany), and TUE (Netherlands). TUE is a participant in the MUTED consortium and brings their considerable expertise to this project, performing the necessary human factors experiments to optimise the MUTED system design.

Various studies, including those of MUTED consortium members, have indicated that stereoscopic pictures are highly appreciated and preferred over monoscopic pictures as long as eyestrain is kept within acceptable bounds. Nevertheless, displaying stereoscopic pictures at high quality with high viewing comfort using a display system that is practical, affordable (assuming mass production), and suitable for professional and home viewing remains a considerable challenge. 3D display design requirements include an image quality reproduction which is likely to be at least comparable to a conventional 2D display, with optimal depth reproduction, and optimal viewing comfort, including minimal crosstalk, and freedom of head and body movement without any tracker devices attached to the body.

Characteristics of 3D display that might negatively affect the viewer experience, such as degradations of the image, reduced viewer freedom of movement and crosstalk, impose constraints on display components such as tracking accuracy, tracker latency, and steering 3D images to the eyes of viewers. Perceptual acceptability guidelines for these impairments will be used to facilitate the design and optimisation of 3D displays.

As an example, early perceptual studies on crosstalk levels revealed a perceptual threshold level of crosstalk at 0.3%. However, subjective acceptability studies performed under the FP5 IST ATTEST project by MUTED consortium members showed that crosstalk levels of 2% are perceptible, but not annoying. Thus, a clear understanding of the perceptual consequences of various display characteristics is of fundamental importance in 3D display optimisation. However, various factors have a large effect in determining the subjective effects of crosstalk, for example, brightness, contrast, stereo disparity and image content. A more complete understanding of these effects is essential if the widespread viewing of stereoscopic images for prolonged periods is to become acceptable without causing, for example, nausea and headaches.

Although a number of 3D display characteristics that may attenuate perceived quality of stereoscopic images are well known (e.g. crosstalk, brightness, picket fence effect, image flipping, puppet theatre effect, etc.), their relative importance (perceptual weighting) is still largely unknown. Thus there is a clear need for a design-relevant and coherent stereoscopic image quality model that will take into account the trade-offs between the various technical parameters, and will allow for subjective weighting of the various attributes. Moreover, subjective methodologies originally defined for monoscopic pictures, e.g. methods defined in recommendation ITU-R BT 500-10, are also adopted to assess stereoscopic images, and are thus limited in their relevance and sensitivity to the particularities of stereoscopic display quality. Well-defined assessment factors, experimental paradigms and stimuli suitable for stereoscopic images are needed to define a standardised subjective performance measure for 3D display systems.

Adequate stereoscopic depth reproduction is obviously important for 3D display design, content production, and evaluation. Perceived stereoscopic depth is a geometrical function of screen parallax, viewing distance, and eye-separation of the observer. As a rule of thumb the ortho-stereoscopic condition is usually recommended. However, recent insights into stereoscopic depth perception for TV applications have shown that a minimal screen parallax (micro-stereopsis) can result in a considerable depth appreciation whilst keeping crosstalk and eyestrain to a minimum. It of considerable practical and scientific value to establish exactly what range of reproducible depth will provide the user with the most satisfying image quality, and produce the least possible eye strain.

Although the display hardware itself may have the capability to display highly appreciated 3D images, if the broadcast content contains errors (vertical disparity, image warping, compression artefacts etc) then this will result in a degraded user experience. Part of the human factors Work package 6 will be to evaluate broadcast images and examine the feasibility for incorporating image processing algorithms within the display hardware that provide image correction to allow the best possible images to be displayed on the 3D display. For example the perceived 3D effects could be exaggerated in certain scenes in order to increase their impact.