Desktop 3D displays vary in their optical design and this results in a significant variation in the way in which
stereo images are physically displayed on different 3D displays. When precise depth judgements need to be made
these differences may become critical to task performance. Applications where this is a particular issue include
medical imaging, geoscience and scientific visualization.
We investigate perceived depth thresholds for four classes of desktop 3D display; full resolution, row interleaved,
column interleaved and colour-column interleaved. Given the same input image resolution we calculate
the physical view resolution for each class of display to geometrically predict its minimum perceived depth
To verify our geometric predictions we present the design of a task where viewers are required to judge
which of two neighboring squares lies in front of the other. We report results from a trial using this task where
participants are randomly asked to judge whether they can perceive one of four levels of image disparity (0,2,4 and
6 pixels) on seven different desktop 3D displays. The results show a strong effect and the task produces reliable
results that are sensitive to display differences. However, we conclude that depth judgement performance cannot
always be predicted from display geometry alone. Other system factors, including software drivers, electronic
interfaces, and individual participant differences must also be considered when choosing a 3D display to make
critical depth judgements.