Rendering, Adaptation and 3D Displays - 3DTV: Processing and Transmission of 3D Video Signals - page 136

1920 1920

1080 1080

L R

960 960 1920

540

L

+

+

1080

540

L R

R

Horizontal downsampling
side-by-side
frame-compatible

Vertical downsampling
top-bottom
frame-compatible

1920

R

L

1080

No downsampling
frame-sequential

R

L

Figure 5.10 3D stereoscopic video display formats [5]

Most display manufacturers prefer to keep the specific pixel/sub-pixel
structure of the shown multiple views a secret, and users therefore are
forced to use proprietary tools in order to be able to display 3D content.
For such displays, developers and researchers are commonly forced to
reverse engineer the sub-pixel structure [6]. The straightforward way of
representing display-specific sub-pixel patterns is to create a texture map,
which has the same size as the underlying display panel, where the view
numbers are represented by R, G, and B. The generated texture in the
renderer is used as a look-up table during the assembly of the final multi-
view image (i.e. scrambling). Figure 5.12 shows a sample display texture
map for the Newsight multi-view display that is capable of displaying eight
views. According to the representation in Figure 5.12, the pixel at location
x

= =

1 (top-left corner) is sampled using views 1, 2, and 3. The red
sub-pixel is sampled from view 1, whereas the green and blue sub-pixels

1, y