More than one way to skin a cat
There is something I've been trying to figure out how to add to the Advantages post, that could be understood by most people why LEDs are an advantage. I'm going to try it here and hopefully work it out before it goes into the other post. If you all understand it then maybe you can help me rephrase it appropriately.
In all systems, the goal is to control the energy reaching the screen at one point relative to another point on the screen. But the energy levels must be more varied than simply full on vs full off. This could be called an analog function; a system that contains significantly more than two states.
But a digital system only knows two states, so for a digital system to produce an analog output it takes advantage of averaging over time and is called Pulse Width Modulation.
So a Full On for 1/9th of the time, then Full Off for the remaining 8/9ths of the time would average out to 11% of Full On (a mostly Off system).
A system like I just described could have 10 analog states (0%, 11%, 22%,... up to 100%) and makes use of a single variable.
LED projectors would have TWO variables to control the amount of energy reaching the screen; they can modulate the mirror and they can modulate the LIGHT iteself!
Why is this a big deal? Well, when combined it allows for some strange timing conbinations.
Lets go back to our 1/9th example and add something else; lets say that during the 1/9th we have the light at 50% intensity and then during the other 8/9ths we have it at Full Off (0% intensity). Our average is 5.5%, not 11%
So if we wanted 11%, how would we do that? There are now two methods:
1.) 1/9th Full On, 8/9ths Full Off
2.) 2/9ths 50% On, 7/9ths Full Off
Lets apply this concept to the 8-bit red channel of a projector. Normally, to display the very dim color of 1/255 red, we would need to flip the mirror on for (1/30 of a second) x (1/3 of that time for red) x (1/255) = ~43.6 microseconds (or uS) of 100% light.
This means that our system has be be able to flip the mirror on and then off again in 44uS to display that 1/255 shade of red (the 1/255 is also called the Least Significant Bit, or LSB, time)
But with the LED, we can use approach number two from above! We know that by halving the Light during the LSB, we are able to double its duration and still get the same amount of energy... with one small kink; since we don't gain any time, we have to make up for that doubled LSB by increasing the length of the system:
(50% light) x (1/30) x (1/3) x (2/256) = 86.8 uS of 50% light (equal to 43.4 uS of 100% light)
Ah-ha! will yell the critics, you've got less energy and now your system isn't accurate! Well, not entirely...
To increase the duration of the LSB I redivided the rest of the time, and this creates a shift. The next bit would be 2/256 at 100% instead of 2/255, then 4/256 at 100% instead of 4/255, and so on until you got to the MSB of 128/256 instead of 128/255
What this means is that Full On (255/255) becomes 255/256 of the maximum energy possible in 256 of 256 periods... but the light is never off in all of that time... we lose 1/256th of the maximum energy possible in that doubled-LSB, because theoretically we could have had 100% instead 50%. But the system runs at our 100% light level definition.
The linearity problem's root is that we were at first saying 1/255th of the energy but then applied that to a 1/256th system (the LSB+1 was 2/256)
To solve the linearity problem on the LSB, the light source is run at 50.195% or (1/255) / (2/256)
OK so I'm trying to point out that there are now two variables and that allows for either relaxed timing or enhancing grayscaling... should I just talk about dynamic range instead? Its difficult for me to teach because the concept is so new to me....