| Author | Message | ||
     alex d Rating: N/A |
I read the the little webpage referenced in the Crosstalk article (http://www.stereo3d.com/bixer.txt) and got inspired. Now it's probably impossible to implement this in software these days, but it seems like it should be possible to rewire the glasses for the same effect with just a few bits of electronics. Basically, the reason that guy was able to eliminate ghosting for page-flipping is simple. Say the monitor just finished drawing an imagine on its screen for the left eye. Right away the glasses switch to the right eye. However, the old left-eye image is still perfectly there! It has started to fade out, especially at the top, and the next image is beginning to be drawn, but the old one is still there. That's, after all, why you get ghosting (at least as long as your glasses can't be faulted). Dave Bixer's solution is to darken both lenses and create a bit of a delay to let the phosphorus cool before opening the next lens. (Another thing that might help is darkening a lens a bit early to allow it to darken before the next image is drawn.) Bixer seems to have achieved zero ghosting with this technique (although he implemented it very aggressively and shut off both lenses for 2/3 of the time). The question, now, is if we can somehow add a bit of electronics to our glasses to achieve the same effect. Ie, to keep the lenses from becoming transparent longer (while at the same time not increasing the time it takes for them to become dark). So... any ideas? If LCDs don't have a preference for the direction of current, then maybe something as simple as connecting one lead to the other with a big capacitor might do the trick. When the circuit goes from opened to closed, the capacitor won't interefere with the LCD darkening, while when the circuit is opened the capacitor will discharge itself by pumping current in the reverse diction. Perhaps what is first needed is a good article on the electrical characteristics of LCDs. If they behave like diodes (with a current preference and a threshold voltage), then things would get much more difficult. | ||
| alex d Rating: N/A |
I haven't found any especially good articles, but I have learned that "The transmission of the LCD as a function of applied voltage is shown in Figure 4.2. There is a threshold behavior for most LCDs and no change in transmission occurs until a threshold voltage, Vth, is reached. Transmission then decreases as the voltage increases until saturation is reached. Threshold voltage is typically 1.5-2.5 volts, and saturation occurs at about 4-5 volts." Voltage is RMS voltage. Therefore, it seems that my capacitor idea should fundamentally work. What's important is what actually happens when the controller cuts off current to the LCDs. If a true open circuit is formed, then my idea should work simply. If instead both leads are grounded or raised to the same voltage or something, then I'd require extra resistors and things would get more complicated. Depending on the resistences in the circuits involved (and of the LCD panel), it may end up working well or poorly. (If the panel has low resistence, then the circuit supplying it with juice will have a high resistence to keep the current low. This will mean that the capacitor would charge slowly, delaying the darkening of the lcd, and then discharge quickly, failing to delay it becoming transparent.) However, I haven't yet received my shutter glasses, and I won't have access to any sort of tools for another two weeks. However, I'll make sure to keep everyone posted when I do try this out. | ||
| Alatar Rating: N/A |
The way I think about this is that each lens should be driven by a square wave, with a duty cycle of slightly under 50% (for the reason you explain), and the signals for each of the two panels should be 180 degrees out of sync with respect to each other. However, the sync relative to the CRT's vsync signal also should be slightly adjustable, partly because of the phosphor persistence you refer to, but also because LCD panels have a lag inherent to them. In other words, if vsync is at zero degrees, LCD on should be at X degrees and LCD off should be at Y degrees; both X and Y being adjusted to compensate for phosphor persistence and LCD panel latency. | ||
| alex_d Rating: N/A |
Yes, precisely. However (contrary to my initial guess), the paper linked to at the bottom indicates that the clear-to-dark period for LCDs is very brief, and shorter than the blanking interval. Therefore, it seems that we can allow LCD on to be at 0 degrees (provided the supply current from the controller is correct and square). http://www.curtin.edu.au/curtin/centre/cmst/publicat/2002-09.pdf So... what can we do? An article for building a controller for SEGA lcd glasses (http://www.tkk.fi/Misc/Electronics/circuits/sega3d.html) says that these glasses need to be powered by an AC current. I'd guess other glasses work the same way. This certainly throws a wrench in the simple capacitor idea. Using the built-in controller in Revelator glasses, which is controlled by a DC voltage, also won't work because the controller is designed for having exactly one glass on and the other off, and the control signal cannot change that. It seems that in order to achieve what we desire, what is required is a redesign of the controller. Anybody up for the challenge? What is clear is that the aforementioned SEGA design won't work. It operates by feeding two unchanging AC currents, 180 degrees out of phase, one to each LCD. The common ground is also an AC signal, which is in phase either with the current feeding the left eye (so that the difference between the two currents is 0 and the LCD panel is clear), or with the right eye (making the left panel dark). This doesn't allow for both panels to be dark. Our design must keep the ground current unchanged and instead independently alter the other two wires. We may utilize the phase-synching approach as used above or use switches (transistors) to turn currents on and off. another SEGA controller based on the same principle: http://www.ee.washington.edu/conselec/CE/sp95reports/mortenso/project.htm |
