http://nikonrumors.com/forum/topic.php?id=1623

TaoTeJared said:
I'm trying to find it again on Sony's site but the light loss is much less than 10% (I believe it was less than 5% but I am trying to find the release with it on it) which does not equate to 1/2 or even 1/3 a stop. All of the light loss hype is just speculation. You would get much more loss with a polarizing filter on.

Again, if the light loss is only 10% to the image sensor it is 90% to the AF sensor. Great that you don't increase photo noise, sucks that you kill low-light AF.

jonnyapple said:
I also thought the former, so maybe I'm not understanding the original comment. It may not even be a half and half splitting of the light, so the numbers could be different. In any case, more than a stop of light lost to the AF system I think we agree on and that's not a good thing.

Ok, let me rephrase and attempt to measure if we are on the same page or not:
It was my understanding that the current system consists of a mirror with transparent (totally unsilvered) segments in the shape of the AF sensors where 100% of the light passes through, bounces off of the secondary mirror behind the one everybody knows and loves, and strikes the sensors below.

You appear to be saying that I am incorrect and that these "windows" in the main mirror are not transparent, but rather semi-silvered and reflect much of the light up to the pentaprism (pentamirror for you with cheap bodies) and the rest passes through to the secondary mirror behind the one everybody knows and loves...

If you are correct, yes, my original "edit" math is clearly wrong and the situation caused by moving from a flippy mirror to a static "translucent" mirror is incalculable until we know what the current transmission percentage is.

I'm trying to find it again on Sony's site but the light loss is much less than 10% (I believe it was less than 5% but I am trying to find the release with it on it) which does not equate to 1/2 or even 1/3 a stop. All of the light loss hype is just speculation. You would get much more loss with a polarizing filter on.

Canon did this many years ago as well as a few others though out the years. I think it is a great solution for shooting video though. I don't think it deserved the Pop-photo camera of the year award though. It is still a sony camera.

Drab said:
"Partially silvered" as in the entire mirror plane isn't silvered (AF windows are transparent) or "Partially silvered" as in the AF windows themselves are translucent?
I thought the former, thus my math based on current AF sensors getting all the light, just all the selected light.

I also thought the former, so maybe I'm not understanding the original comment. It may not even be a half and half splitting of the light, so the numbers could be different. In any case, more than a stop of light lost to the AF system I think we agree on and that's not a good thing.

Drab said:
I like how you always bring it back to FX AF frame coverage. ;)

Sketch available on demand, but doesn't that theory just preclude AF points near the top and bottom edges of the FX frame, and not the left and right edges?

I always come back to it because I've always naturally come back to things I don't understand. You make a great point about the horizontal vs. vertical coverage and it now makes sense to me why there would be more horizontal coverage with the current design.

I like this x-ray image of the D700 from kenrockwell near the bottom of this page:
http://www.kenrockwell.com/nikon/d700.htm

jonnyapple said:
True, Drab (edit: I don't think your edit is true, because current AF sensors get half the light. You need to compare 1/6th with 1/2, which is only a factor of 3 or about 1.7 stops—still nothing to sneeze at). It's also incompatible with the current Nikon AF and metering system because, currently, Nikon also has a partially silvered mirror with another mirror behind it so that light can get through it to the AF sensor,

"Partially silvered" as in the entire mirror plane isn't silvered (AF windows are transparent) or "Partially silvered" as in the AF windows themselves are translucent?
I thought the former, thus my math based on current AF sensors getting all the light, just all the selected light.

But, yea, 1.7 stops is still atrocious. We all know how poorly a f/5ish maximum aperture zoom focuses in dim light.

jonnyapple said:
I hadn't thought about it before but the back mirror, which is at a right angle (ish?) to the front mirror, could actually be one of the main limitations on frame coverage for the AF sensor—you can't have a mirror that's the same size because you'd run into the shutter plane.

I like how you always bring it back to FX AF frame coverage. ;)

Sketch available on demand, but doesn't that theory just preclude AF points near the top and bottom edges of the FX frame, and not the left and right edges?

True, Drab (edit: I don't think your edit is true, because current AF sensors get half the light. You need to compare 1/6th with 1/2, which is only a factor of 3 or about 1.7 stops—still nothing to sneeze at). It's also incompatible with the current Nikon AF and metering system because, currently, Nikon also has a partially silvered mirror with another mirror behind it so that light can get through it to the AF sensor, which is below the mirror, while still getting to the metering sensor, which is above the mirror in the prism housing.

I hadn't thought about it before but the back mirror, which is at a right angle (ish?) to the front mirror, could actually be one of the main limitations on frame coverage for the AF sensor—you can't have a mirror that's the same size because you'd run into the shutter plane.

monty11 said:
But you are losing only one stop in terms of ISO, so the result isn't that bad considering how low the ISO noise is at the moment.

If one stop doesn't matter than the D3s is a waste over the D3.

Paperman said:
1/3 to 1/2 stop apparently , corrected by ISO boost .

The downside is the more light they give the imaging sensor the less light they give the AF sensor, impacting AF performance.
There isn't a free lunch here.

EDIT:

And (check my math), but if only 1/3 of a stop is lost at the imaging sensor, that means the AF sensor is only getting 1/6th the light it would get in a traditional SLR system. What's that, 2.5 stops loss for the AF system? A 2.8 pro zoom has the low-light AF hunting of a PoS f/6.7?

1/3 to 1/2 stop apparently , corrected by ISO boost .

But you are losing only one stop in terms of ISO, so the result isn't that bad considering how low the ISO noise is at the moment.

IndyGeoff said:I wonder what the down side is.

You've split the light so now less light hits the sensor, so for any given sensor you will have more noise.

IndyGeoff said:
SONY A55 TRANSLUCENT MIRROR

Sounds very interesting. I wonder what the down side is.

We have the sensor to keep clean ; A55 users have the sensor and the mirror to keep clean :-) ...

PP has chosen it as the camera of the year but also expressed concern about the translucent mirror going dark in time .( I don't think Sony will worry much about that having introduced 12 ( yes, twelve ) entry level DSLRs in 15 months ! )

Slight ghosting due to the pellicle mirror was also reported in Dpreview.

apologies for not using the search first. can kill post.

SONY A55 TRANSLUCENT MIRROR

In a regular S.L.R. camera (single-lens reflex — those big black pro cameras), light enters the lens, hits a mirror and is bounced up to your eye and, simultaneously, onto a focusing sensor. Unfortunately, when you take the photo, the mirror has to flip out of the way so that the light falls on the image sensor (the “film”). At that point, the camera can’t focus. That’s why most S.L.R.’s can’t change focus during burst-mode shots, or while filming video.

Sony’s A55 camera ($850) solves that problem by using a translucent mirror. It splits light between the focusing sensor and the image sensor. The mirror never moves, so the autofocus never goes blind. The camera can take 10 shots a second, refocusing all the way — no other camera can do that — and change focus as you pan or zoom, gorgeously and cinematically. No wonder this was Popular Photography’s camera of the year.

Sounds very interesting. I wonder what the down side is.