Reader Ryan has pointed me at a slightly old (in 3D talk, August 2011 is old) discussion of Active vs Passive 3D technology in Audioholics. Many of the points it makes are well founded, but it also demonstrates a misunderstanding of the brightness issue. With regard to active 3D it says:
… but because the shutter effect literally closes off light to one eye at a time, the use of active 3D glasses is said to cut brightness significantly. With a drop in brightness comes less fine detail in coloration and shadows.
But this is actually misleading. In fact, in a theoretically perfect passive 3D system and a theoretically perfect active 3D system, all else being equal, the brightness would be identical for both.
In such a perfect active system, for sure the brightness would be half in 3D of what it is in 2D because only one eye is being exposed at a time, while the other is being blocked.
But in a perfect passive system, again the brightness would be half in 3D of what it is in 2D. The reason? In 3D the left eye can only see half of the display lines and the right eye can only see the other half! (This is a separate issue to the resolution question.)
In both cases — and again, I’m talking only of theoretical perfection here — you get half the brightness that you do in 2D because both systems are filtering out half the brightness: temporally in the case of active, spatially in the case of passive.
So what could account for real world active systems being perceived as dimmer than passive ones?
In part, it is by no means anywhere near as clear as it was before because active 3D technology has improved, even for LCD TVs. But to the extent that it existed before, and continues to exist, dimness unique to active 3D technology is to do with switching.
I am going to simplify matters by omitting discussion of scanning backlights in LCD TVs and subfield pulsing in plasma. Indeed, I’ll leave out plasma completely at this point.
Now we know that it takes a certain amount of time to switch an LCD pixel from one state to another. Figures of 3, 4, 5, 6 milliseconds seem to be commonly claimed ‘grey to grey’ pixel response times. While milliseconds seem like a very short time, it must be remembered that a movie frame of a 24fps Blu-ray lasts for only 41.7 milliseconds, so that switching time is something like 10% of the total frame display time or less.
So to some extent LCD displays can smear detail over time when said detail is moving. This is partly addressed by such things as scanning backlights (they can blank briefly during the transition, for example), but isn’t that much of an issue in regular viewing for a least two reasons. First, most moving content is shot with an exposure time that in itself produces some smearing in the direction of the motion. So what does a little more smearing matter? Second, we aren’t that good at perceiving fine detail on moving objects anyway, so again we don’t perceive the smearing too much.
But the job of an active LCD TV has different demands. It needs to display the right eye information, and only the right eye information. Then it has to make that go away so it can display the left eye information and only the left eye information.
If any of the right eye stuff is still on the display (as the pixels are in a state of change) when the active glasses have blacked out the right lens and opened up the left lens, you will see crosstalk.
How to deal with this? Simple: when it’s time to switch from the right eye to the left eye, blank out the right eye lens … and leave the left eye lens blank as well! Leave it blank until the last vestiges of the right eye picture have completely gone, and only then open the left eye lens.
And that’s why active systems are darker than passive systems: each lens spends more than half the time blanked out. If LCD switching was instantaneous then there would be no need to have both lenses blanked at the same time, so brightness would be 50% of 2D, the same as passive systems.
But they aren’t, of course.
So how long do both eyes have to be blanked? Bearing in mind that the maximum display time for a frame sequential active display with Blu-ray 3D content is just 20.8 milliseconds?
That I do not know. My guess is that it would be in the range of 2 to 8 milliseconds. What I do know is that dimming is a sufficiently important issue that TV makers attempt to strike a balance between a long double-blank time to eliminate crosstalk and a short double-blank time to maximise brightness. And that is why active* 3D LCD TVs (and LCoS and LCD projectors) always have some level of ghosting, because the hit on brightness would be too great to keep both lenses blanked until it was completely gone.
All this reminds me that I’ve covered some of this territory, and especially active vs passive when it comes to ghosting, in an earlier post: 3D Crosstalk is Always and Everywhere a Timing Phenomenon.
(Note: when I say that 3D might have 50% of the brightness of 2D, I am ignoring the filtering effect of the transparent states or sections of the eyewear. In reality it must always be less than 50% for both forms.)