3D image reveals a novel perceptual phenomenon

Creating 3D images for future media and technologies has long been a subject of active study.¹ Understanding how the human visual system copes with 3D problems is an important part of this research. We previously reported a depth-fused 3D (DFD) display based on the discovery of a new perceptual phenomenon induced by human binocular vision.² The DFD system comprises two 2D displays arranged at different depths. The images on the screens differ only in their luminance (intensity). When they are viewed as completely overlapping, the images are seen as a single image located between the two displays. The perceived depth of the fused image depends on the luminance ratio of the two images. In other words, an apparent 3D image can be produced anywhere in the space between the front and back displays by changing the luminance ratio at each pixel.

Here we describe a new type of DFD display based on stereoscopic rather than 2D screens (see Figure 1). We wished to find out whether two 3D stereoscopic images shown at different depths would elicit the same perception as a conventional DFD display. We also wanted to understand what factors influence the perceptual process.

Figure 1. Schematic representation of a new depth-fused 3D display system. Z_F indicate the perceived depth of 3D images on the front stereoscopic screen alone. Similarly, Z_B indicates the perceived depth of 3D images on the back screen, when no image is displayed on the front screen.

Figure 2. Apparent depths of fused images. The x-axis indicates the ratio of the test pattern luminance of the front stereoscopic screen to the sum of the luminances of the patterns on both displays. The empty circles, triangles, and squares show the average depth reported by six observers when the depth perceived by binocular disparity is at the position indicated by the black circles, triangles, and squares.

Subjective tests of this display revealed that two similarly shaped 3D images on stereoscopic displays did indeed blend when they overlapped. Figure 2 shows the perceived depth of the fused image obtained in the test. The impression turns out to depend on both the luminance ratio of the images and the different vantage points of the viewer's two eyes (binocular disparity).

These factors are decoupled in the 3D display, whereas in a standard DFD system they are not. This decoupling separates disparity-related depth from focus-related depth, since the focal cue continues to determine the actual distance to the screen. We concluded that DFD perception is dominated by the effects of binocular disparity and image intensity, that is, the influence of the depth of focus is much weaker. As a next step, we intend to clarify what the human visual system actually focuses on when looking at a fused 3D image.