kilian@poplar.cray.com (Alan Kilian) (06/05/91)
> lance@motcsd.csd.mot.com (lance.norskog) Asks: > > 1) You reported that your Sega Goggles "don't go very dark". > Do you have an extinction ratio rating for them? No, But I will measure them tonight and get a reading to you tomorrow. > 2) > The naive scheme is to switch between states based on the vertical > retrace. Another thing is that a vertical retrace does not give you a left/right image identifier and it is up to the user to choose the proper "Phase" for the glasses. Also it is necessary to have a display system capable of insuring alternation of left/right images every other frame time. Most channel attached framebuffers can not do this. (A channel is a wire connecting the supercomputer to an external device like a disk drive or framebuffer). I have developed a system using an optical detector and an area in each frame dedicated to the left/right image identifier. This system allows a device to remain in synchrony with images even if they are not alternating every frame. It can be used for projection movies, regular computer CRTs and video tape displays on conventional televisions. It is Patent Pending though so no commercial applications can use it without paying something. > It seems that you may want to decouple the square wave > to each lens from being alternate portions of one square wave, and > overlap them a bit, initiating the clear phase just before the > vertical retrace. Hmmm let me think about that. Let's get some numbers in here. (Shuffle shuffle) O.K. 120 frames per second. 1.3ms from full phosphor output to zero output. 2ms to switch the LCD from one state to another. (On ->Off or Off -> On) How about 7ms for each frame to be drawn every 8ms. Now if you switch at the vertical retrace time it will be about .5ms after the last raster line was drawn and it will be 1ms from full extinction. By the time the glasses get to their other states (1.3ms) The first 14% of the raster lines are drawn and will be seen partly by the wrong eye. Also since the switching time is greater than the phosphor extinction time we don't have to worry about the lower part of the screen. So you are right, we want to start switching earlier. How much earlier? Well if ST is the SwitchingTime and ET is the ExtinctionTime and IRG is the Inter Raster Gap ST=2ms, ET=1.3ms, IRG=1ms We know we don't want to start switching any earlier then ST before the last raster line gets drawn or we won't be able to see it. So how about .5*ST before the last raster line? I think that that would be O.K. We also don't want to start switching any later than (LR+IRG)-ST otherwise the first raster line will be seen in the wrong line. So how does this new system fair? We now start to switch and get fully switched before the first raster line, but AFTER the last raster line has "gone out". Pretty good huh? I am going to move my sensor to the bottom of the screen and recompute the timing variables for this new system soon. Thanks. > Perhaps the off phase should be initiated right around the > bottom of the screen, so as to cut off dying phosphors because > the colors decay at different rates. That's exactly what I come up with going through the numbers. Good intuition. > 3) > The right solution for large screens (> 500 lines) would seem to be > bifocal shutters. The top half matches the top half of the screen, > and the bottom shows the bottom of the screen. This gives tighter > control in matching the scan rate of the screen. I don't really understand this system. Do these shutters go on my face or on the CRTs face? If they go on my face then I don't see how I can control where the shutters boundary lines up with the raster lines. I COULD rotate my head 90 degrees and look at the screen sideways if I wanted to. So I have to assume that they are in front of the CRTs face. Now why would two independently controlled shutters help? I don't know but I'll think about it. Hmmmmmmmmm. O.K. I've got it. After the first half of the screen is drawn and goes extinct you can switch the top half of the shutter to opaque and not worry about it. Whoops I just realised that these things don't go opaque, they switch left/right So I should say: When drawing a left eye image after the first half of the screen is drawn and goes extinct, the upper half of the shutter can be switched to the right eye mode in preparation of the first raster line. Then just before the end of the left eye image you can begin to switch the lower half of the shutter to the right eye mode as we computer above. Pretty cool. I think that this would work just fine. > 4) > A hacker at Vision Research Group (another LCD shutter vendor) > told me that the control waveforms should come out of a ROM > instead of an analog circuit; this gives better performance somehow. I don't buy it. Computer people are terribly intimidated by analog electronics and seem to go to digital ROM things far too often. An analog circuit can have variable resistors in them to tune for a particular application and be much more versatile. A ROM based waveform thing needs a new ROM. Boo. I really don't like burning ROMs. > The next generation of LCD shutter gear should be designed based on > these conclusions: Separate the duty cycle for the two lenses. > Adjust the duty cycles of the lenses for the switching times > of the lens and the phosphor decay times of the colors of the monitor. > For big-screen work, try bifocals. Good conclusions. I'll do just that. Thanks, -Alan "ROMs, we don't need no steenkeen ROMs!" Kilian -Alan Kilian kilian@cray.com 612.683.5499 Cray Research, Inc. | "The Fragile X Syndrome may me the most 655 F Lone Oak Drive | frequent cause of inherited mental Eagan MN, 55121 | retardation". Science 24-May-1991 PP1097