91-05/Tek3d.monitor Here is a quick evaluation of the Tektronics SGS610 Stereoscopic System. Note that I do not give any warranty to the accuracy of this info. Basic Principle: The idea is to flip rapidly between images created for the left and right eye. These images should be created by taking a snapshot of the world and then translating ~6.5cm left (or right...arbitrary depending on which eye you are rendering for) recentering the view frame (NOT rotating the camera) and taking another snapshot. These images are then flashed on the screen synchronously with the flipping of a shutter. This shutter (in the Tektronics case) is a LCD screen which fits over the computer screen. The shutter switches between acting as a right circularly polarized filter and a left circularly polarized filter at rates of up to 120 Hz (60 Hz per eye). The user wears passive glass which are right and left circularly polarized over the right and left eye respectively. Note that these glasses are passive and look like normal sunglasses. Thus they are light and non-electical with no tethers. Provided the images are in sync with the shutter and the appropriate image is displayed for each case, the user should get the proper effect: the left image only seen by the left eye and the right image only seen by the right eye. More information on circular polarized filter can be found in most advanced physics textbooks dealing with light. What's included (paraphrased from 1988 User Manual): 19" (610), 16" (410), or 12" (310) liquid crystal Stereoscopic Modulator which is attached to the user-supplied monitor Stereoscopic Modulator Driver with AC power supply Four pairs of viewing glasses Modulator Interconnect Cable Velcro Mounting strips (used for attaching modulator to display) User's Manual What you do: First you must have a graphics board which can switch images at least as fast as 60 Hz (30 each eye). In my case this is a Sun TAAC accelerator in a Sun 4/360 (the images are ~500x500). These boards synch using the synch pulse sent to the monitor. The modulator driver also sits on this synch line so that it can keep the modulator in synch. The velcro strips are used to hold the modulator on the computer screen. You render the appropriate views as described above, start the stereo mode on the graphics board, plug in the modulator driver, put on your glasses, and suddenly you have 3D! Note that depth reversal may occur due to the right image being given to the left eye, etc. This can be fixed by flipping the depth reversal switch on the modulator driver. Also, the image will appear to follow you. This is due to the lack of motion parallax in the system. While you are moving, the same images are being presented to your eyes. Thus, the objects seem to follow you. More Info: There are six different modes for giving synch pulses to the driver. Control inputs are TTL levels; composite synch levles are >= .2V neg. synch and (<5V p-p video and sync). These are vector direct (from one input (BNC connector) : on = right, off = left), vector latched (first true on input 1 latches right eye on, first true on input 2 latches left eye on), vector gated (at true edge of input 1, the data at input 2 will be used....if it is true then right eye else left eye), raster frame direct (raster mode...on true edge from input 1, the right eye is turned on, the right eye is turned of and the left eye on at the appropriate interval afterwards...delays for modulator switching and phosphor decays of 1.5ms or less are included), raster composite sync (hard to explain...right thing for a Sun monitor model HM-4119-S-AA-O however...again compensates for phosphor decay and switching), and raster field direct (again hard to describe but compensates for phosphor decay and switching). We use raster composite sync. There are also options for termination or feed through of sync input and 30 or 60 Hz rates. We use the 30 Hz rate without termination or feedthrough (last two did not effect performance). Interesting Manufacturer's Specs (again from 1988 User's Manual): Warmup time 60 sec MAX (ours about 10-20 secs) Right eye turn-on time 0.35 mS MAX (switching from left to right) Left eye turn-on time 3.2 mS MAX (switching from right to left) Average light transmission 12% (????? seemed much better than that) Ave extinction ration (on image/off image) left red 14/1 14/1 green 9/1 10/1 blue 5/1 8/1 right red 20/1 20/1 green 15/1 15/1 blue 10/1 10/1 The display used needs to have fast phosphors (~1.5 msec for decay) for this system to work well. Equipment Used: This is primarily a rehash, but: Sun 4/360 with TAAC graphics accelerator board Sun monitor HM-4119-S-AA-O Tektronics SGS 610 19" Sterescopic 3D Display Kit, 120/60Hz (running in 60Hz mode) Software Environment: ThingWorld 3D Modeling System (A. Pentland and a host of others in the Vision & Modeling Group, MIT Media Lab). Runs primarily under X. Evaluation: The binocular disparity provided by this system causes a rather striking depth effect (as exclaimed by several viewers). Unfortunately, this effect was limited to when only red objects were presented (white and green objects were also tried, a black background was used in all cases). Otherwise, ghosting was prevalent. Ghosting is when the image from one view persists on the screen while the other eye is being addressed. The advantage of red objects is probably due to the good extinction ratio of red as opposed to green and blue as shown in the manufacturer's specs. There are several ways to address this issue. The first is to get a display with faster phosphor decay. It is possible that the display used has a decay rate of > 1.5msec. Another solution may be to increase the frame rate. If the manufacturer's specs are correct, this should reduce the effect somewhat. Unfortunately, our graphics board is not set up for the higher speed. Another solution may be to adjust the timing of the presentation of the views. However, this may be difficult depending if internal adjustments can be made to the modulator driver or the accelerator board. Another drawback was the flicker observed in our system. Again, the higher frame rate could be used to reduce this effect. There are several major advantages that have been noticed with this system. They are: 1) Portability of the stereoscopic system to other displays and systems (as long as the graphics and sync requirements are met). 2) Passive glasses. This is a major advantage. a) There are no tethers to the system. The glasses feel and look (unless viewed through another pair) like normal sunglasses. b) Multiple users can view a screen at once (although they will get different effects from not all being at the optimal viewing distance). c) The user can tilt his head in any direction and still receive left and right views due to the circular polarization method. 3) Lack of mechanical parts (except for the LCD crystals) I can think of one other inherit disadvantage besides the ones already given. This is the interference the modulator can generate. When using the system with 2 Polhemi, strong interference occured. However, this was rectified by simply moving the Polhemi sources farther from the modulator. In the next few weeks I will be experimenting more with this system. One of my goals is to reduce ghosting. If anyone is interested in the results or clarification of the above, I can be reached at testarne@media-lab.media.mit.edu much quicker than the athena account. Also, my S.B. thesis (for which this equipment was used) gives a brief overview of the various stereoscopic and 3D imaging methods available from the Wheatstone stereoscope through holography if anyone is interested. Standard Disclaimers: Note that I do not work for Tektronics and do not give any warranty for my information. Also, while I am affiliated with the Vision & Modeling Group, MIT Media Lab, my opinions and ideas do not necessarily reflect those of the group or lab. Thad Starner