Stereo display technologies like those described in the previous section are able to present different images to each of the viewer’s eyes. How exactly the viewer perceives the displayed scene (in terms of depth and 3D shape) depend on how these images differ and how they are looked at. Moreover, when looking at a point in the real world two responses are triggered in the eye related to depth. On the one hand, the eyes converge to align with the target point in a movement called vergence. On the other hand, the eye’s lens changes shape to focus a sharp image of the target object on the retina, a phenomenon called accommoda- tion. An additional response, pupil dilation or contraction will not be taken into account in this work as it is not as relevant to depth percep- tion. A thorough explanation of the structure of the human eye and its behaviour can be found in Gross et al. [36].
Despite their differences, an eye can be compared to a camera having as main elements a variable aperture (the iris), a focusing element (the lens) and an image projection surface (the sensor or the retina). A typical lens is characterized by its focal length (usually expressed in millimeters) or by its reciprocal, the optical power or refractive power (usually expressed in diopters, 1 D = 1 m−1). Using optical power has the benefit that the power of several consecutive lenses is approximately the sum of their individual powers.
Accommodation is an effect that occurs in each eye even if the other is covered. When looking at a distant object, the eye’s lens is relaxed and its optical power plus the power of the cornea focus its incoming parallel rays into a point on the retina (see Figure 2.3). When looking at a near point, the eye’s lens has to be deformed and thickened to focus rays into a single point. The accommodation optical power required for best focus at each moment is the accommodation demand. The eye’s physiological response is the accommodation, the change in optical power from distant viewing to a shorter distance viewing. It may not be
equal to the demand if this is beyond of the limit of a given eye. This demand can be expressed as an accommodation distance (the distance to the object looked at).
A
A
da
lens
∞
Figure 2.3: Schematic view of accommodation: looking at a distant object (top) the lens is relaxed; looking at a near object (bottom) the lens is made thicker to accommodate (focus)
At the same time, when one looks at an object binocularly, the eyes align with it in order to project the target point on the fovea (the central vision point on the retina). This is schematially depicted in Figure 2.4. Typical stereo displays can trigger the eye vergence response given the disparities in the left and right presented images. But, as the display is physically at a fixed distance, blur-induced accommodation demand remains constant. Actually, the control of the vergence and accommo- dation responses are not independent, each controlled by retinal image disparity and sharpness, respectively. There are interdependencies so that image sharpness/blurriness can also affect vergence and stereo dis-
A
A
A
dc
Figure 2.4: Schematic view of the convergence needed to fixate an object at a given distane.
parity can affect accommodation. The details of these complex inte- ractions are out of the scope of this thesis. Detailed descriptions of these mechanisms can be found in Schor [82], Templin et al. [93].
Figure 2.5 depicts a pair of eyes watching a stereo screen displaying a small object (letter ‘A’) with positive parallax (the letter appears behind the screen). The viewer’s eyes align with their respective images of the object and thus converge at a point behind the screen. The distance to the screen da is where the eyes should accommodate to have a sharp
vision of the object, and that is different to the convergence distance dc.
However, as Inoue and Ohzu [41] found, the disparity in a stereo screen affects accommodation so that the eyes will try to accommodate at the convergence distance dc. In this situation, the image of the object
will be blurry because the eyes do not focus at the correct distance where the screen is located. The difference between daand dccreates a conflict
A
A A dc daA
A
display Apparent position Image for left eye Image for right eyeFigure 2.5: Stereo vision of a stereo display.
known as the vergence-accommodation mismatch.
In the case of viewing a virtual scene in an HMD something similar happens. See Figure 2.6. Here each eye sees a separate screen magnified by optics (one or more lenses), so the distance at which visual axes converge when looking at the displayed letter is not as straightforward to find. The paths these axes follow depend on the position of the displays with respect to the eyes and on the optical power of the lenses. Accommodation distance is also not geometrically obvious. It depends on the optical power of the optics used and the distance between the optics and each screen. This accommodation distance is fixed because those variables (lens optical power and display distances) do not change. The mismatch thus appears in HMDs as well.
A
A A dc A A Left microdisplay Right microdisplay Apparent positionFigure 2.6: Stereo vision in an HMD