Flaw of Time-separation
Dichoptic Presentation – Simulated Pulfrich Pendulum
as an Example
* 2014 International Strabismological Association (ISA) meeting Kyoto *
Ai-Hou Wang
ABSTRACT
Purpose:
Binocular function
tests are designed with dichoptic presentation on computer screens. There are
two mechanisms for 3D screens. Time-separation applies shutter goggles to view
alternate frames with either eye. Space-separation applies circular-polaroid
goggles to view odd and even lines with either eye. We designed simulated Pulfrich pendulum with either method to find the one with
time-separation screen was unreadable.
Methods: Simulated Pulfrich
pendulum exhibited as that either eye viewed a horizontally sine-wave
moving circle. With a lag between these two circles, people would see a fused circle
moving in depth. With time-separation 3D screen, the odd frames exhibited circles
for right eye and even frames exhibited circles for left eye.
Results: As contrary to space-separation screen, you
would not get the depth movement on time-separation screen. The reason was
that, despite you designed frames (1 and 2), frames (3 and 4)….as
stereo-pairs, frames (2 and 3), frames (4 and 5)… exhibited as equally strong
fusible pairs. These unwanted stereo-pairs exhibited another path at a
different depth. It would disturb the pendulum path you would like to show and
made the whole thing an ambiguous or rivalrous 3D
picture.
Conclusion: We suggest space-separation rather than
time-separation screen for dichoptic presentation, especially for dynamic
binocular function experiments.
KEYWORDS:
Dichoptic presentation, 3D presentation, Space separation, Time separation, Pulfrich pendulum
Introduction
3D screen is
useful for both entertainment and research work of binocular vision. To present
separate images for two eyes, there are two mechanisms, time-separation and
space-separation. Time-separation applies shutter goggles which is synchronized
with the screen to view alternate frames with either eye. Space-separation
applies circular-polaroid goggles to view odd and even lines with either eye.
Nintendo applied time-separation screen for 3D games as early as in 1980’s.
Wang AH, Jan ZH, Yu JJH, Lin LLK Binocular
dissociation with interactive graphics on personal computer and stereoviewer. Images of the Twenty-First Century.
Proceedings of the Annual International Conference of the IEEE Engineering in
Medicine and Biology Society 1989; 1680-1681.
ASUS had produced 3D screens with either mechanism (see
figure), but both had been withdrawn from the market.
We designed simulated Pulfrich
pendulum with either method and found that under certain condition the one with
time-separation screen was unreadable. We analysed its mechanism of dichoptic
presentation and found a fundamental flaw of time-separation 3D screen.
Methods
Simulated Pulfrich pendulum
was displayed as red/blue anaglyph. Right eye viewed a blue circle
and left eye viewed a red circle, both move horizontally as sine-wave pendulum.
There was a lag between the blue and red circles. With space-separation screen
the stereo-pair was shown on the same frame; with time-separation screen the
stereo-pair was shown on consecutive frames (see figure).
We prepared 60 consecutive stereo-pairs in a pendulum
cycle. In space-separation screen there’re 60 frames while in time-separation
screen there’re 120 frames in a cycle. As these frames were displayed in
forward or backward time sequence the fused circle would look like going in
depth along a clockwise (CW) or couterclockwise (CCW)
path in the space (see figure).
CW and CCW path were designed to appear randomly. This
served as a 2-alternative-forced-choice (2-AFC) test of the subject’s binocular
depth sensation.
Results
The test designed on
space-separation 3D screen worked fine but on time-separation 3D screen,
sometimes you would get ambiguous depth movement.
The reason was that, despite you intended to design
frames (1 and 2), frames (3 and 4)… as stereo-pairs
(see figure, green stereo-pairs), frames (2 and 3), frames (4 and 5) etc.
exhibited as equally strong fusible stereo-pairs (see figure, red
stereo-pairs). These unwanted stereo-pairs displayed as a different path in
depth and would disturb the pendulum path you intended to show. The whole thing
thus became an ambiguous or rivalrous 3D picture.
discussion
Time-separation 3D screen has the flaw that despite we
would like to design frames (1 and 2), (3 and 4), (5 and 6)…
as stereo pairs, physically frames (2 and 3), frames (4 and 5), frames (6 and
7)… have the same potential to be seen as stereo pairs. This might sometimes
interfere profoundly with our binocular vision experiments.
For 3D movies on blue-ray disc (BD) or television
program time-separation 3D screen has little problem because the pictures did
not change fast. But when applying time-separation 3D screen to dichoptic
presentation for research work of binocular vision you should be utmost
careful.
In our Pulfrich pendulum
experiment, in the condition of Delay = 0.5, frame (1 and 2) and frame
(2 and 3) were equally strong stereo-pairs. The former was crossed-disparity and the path thus came out in front; the latter
was uncrossed-disparity and the path thus went recessed behind.
In the condition of Delay = 0.8, the distance
of the circles in frame (2 and 3) was closer than that in frame (1 and 2) and
was thus easier to fuse. And the subject tended to see a path going in reversed
direction (CW vs. CCW) as what you would expect.
Following the same reason, in the condition of Delay
= -0.3 there won’t be problem of CW vs. CCW ambiguity.
But we don’t think this kind of ambiguity aroused by
time-separation 3D screen should reasonably be involved in our design of
binocular vision experiment.
The flicker fusion frequency (FFF) is 40-50Hz for each
eye, and 60Hz is just enough to avoid flickering for space-separation 3D
screen. Time-separation 3D screen needs doubled frame rate as space-separation
3D screen and thus reaches 120Hz.
Engineers may suggest to insert
a blank frame (see figure) to enhance the weight of frames (1 and 2) rather
than frames (2 and 3) as a stereo pair. This requires even higher frame rate
(180Hz) and this higher technology will certainly
cause higher price of this 3D screen.
On the other hand, we would now consider some
drawbacks of the space-separation 3D screen. There are threefold –
(1) Lower vertical resolution – For high definition
(HD) screen there are 1080 horizontal lines. These are equally shared by right
and left eyes and each eye sees only 540 lines.
(2) One line vertical
disparity between right and left eyes – 23-inch 16:9 high-definition (HD)
screen subtends 50.92 × 28.64 cm. At 50cm distance, each line subtends 1.82
min-of-arc. This is the vertical disparity between the images seen by right and
left eyes, and it is usually tolerable by vertical fusion of human brain.
(3) For independent visual stimuli for right and left
eyes it will be more difficult to program on space-separation 3D screen – You have to combine what seen by right eye and left eye into a
single frame in every instant. For example, designing gratings of different
spatial frequency, different speed and different drifting direction for right
and left eye, you just put them into alternate frames on time-separation 3D
screen. But on space-separation 3D screen you have to
combine them into a single image for every instant. There may be too many
combinations and would take you much more efforts to accomplish that.
conclusion
We suggest space-separation rather than
time-separation 3D screen for dichoptic presentation in vision research,
especially for dynamic binocular function test. The following table compares
space-separation vs. time-separation 3D screen.
