[Leica] Digital cameras with large aperture lenses

[henningw at archiphoto.com (Henning Wulff)]

At 12:21 PM -0800 3/19/05, Adam Bridge wrote:
>Okay - I STILL don't understand what you say is happening.
>
>I can understand that less light may be captured by the microlenses
>over the sensor elements located at the edges of the sensors because
>the lenses, I suppose, are designed for light arriving perpendicular
>to the plane of the sensor when, in fact, they are not.
>
>BUT  how does this affect what's going on in the middle of the sensor?
>It doesn't make any sense to me at all. If you open from f2 to f1.4
>you should receive twice the number of photons arriving at the
>micro-lens (since the lens works identically as that for film up to
>this point.)

First, think of the image sensors being at the bottom of a small 
well, so that it has trouble seeing anything except what is in a 
narrow angle directly above them. The microlenses try to alleviate 
this by providing a wider angle of view, but their efficiency falls 
off the greater the angle of the incident light off the perpendicular.

Now think of a lens with a large aperture, that has its rear nodal 
point close to the censor. For the sake of this argument, consider 
the rear nodal point as the center of the rear element, and the rear 
nodal plane as being the cross section of the rear lens parallel to 
the sensor. The diameter of the rear element is the same as the 
diagonal of the sensor in this lens. Now: a ray that hits the very 
corner of the sensor, coming from the center of the rear nodal point, 
arrives at an angle say, 45 degrees from the perpendicular. As it has 
to make a 45 degree turn at the entrance of the sensor 'well' to make 
it down through the filters and into the sensor, a lot of light is 
lost and vignetting results, right?

Now, imagine that you are using the same lens wide open to shoot 
something, and therefore some of the rays go through the middle of 
the lens and hit the central sensor well (middle pixel of the image) 
head on. Optimal usage of the light. However, with a very fast lens 
wide open, over _half_ of the light comes from the outer 1/3 of the 
lens's diameter (you can check the geometry). This means that a lot 
of the light is expected to go through the outer zone of the lens, 
make a 40 to 45 degree turn through refraction of the lens, and then 
enter that central sensor well site by making another 40 to 45 degree 
turn, which it finds just as difficult to do as the previous ray 
trying to get to the corners. The sensor can't make very efficient 
use of this light, and instead of getting one full additional stop's 
worth of light, it only gets 50% or even less of that.

I hope I'm making this clearer, and not muddier. :-)

Anyway, it's mostly just geometry, and if you draw it out on paper, 
it might become clearer. Remember, it's the rays that have to make a 
sharp turn that have trouble, and in the end don't contribute much to 
the light gathering. Rays come from the full rear nodal plane in a 
real lens, and the closer the rear nodal plane gets to the sensor, 
and the larger the rear nodal plane is, the more rays have to make 
sharp turns to get into the sensor wells.

In so-called 'telecentric' lenses, or specifically the ones that are 
closer to the telecentric ideal in image space, the light rays are 
all more parallel, and as far as the sensor is concerned, come from a 
greater distance (a rear nodal point and plane far from the sensor) 
so that sharp angles are no longer involved at the sensor plane. They 
rays all head 'straight in', and are no longer required to make a 
sharp turn.

>So you are saying that something non-linear is happening at this
>point. What would this be? Is the reflection at the air/microlens
>interface involved? The microlens/sensor interface? It's about
>photons, right? I can't believe we've dropped into the quantum realm
>where QED explainations are required in order to understand the
>interactions. But I sure could be wrong.
>
>anyway - I'd like to understand the physics and so far it's all
>hand-waving and assertions.
>
>There must be a way to do some science here - but I think the camera
>software, even in RAW mode, tends to cover up any effects that might
>be happening.
>
>I'm really curious and dumbfounded - I'm not doing this for the sake
>of argument.
>
>Adam
>_______________________________________________
>Leica Users Group.
>See http://leica-users.org/mailman/listinfo/lug for more information


-- 
    *            Henning J. Wulff
   /|\      Wulff Photography & Design
  /###\   mailto:henningw@archiphoto.com
  |[ ]|     http://www.archiphoto.com