Olympus E-3 User Manual page 68

One vital fact, however, is that the system needs to be able to
analyse the differences in the two side's light patterns. If there
are no differences – as in when the camera's focus point is
seeing only one shade or colour without any contrasting line
falling within the area of the focus point - both sides will "look"
the same to the computer and it will either sit and sulk (i.e. do
nothing) or whiz the lens to and fro in order to try and "hunt
down" some difference in the phases.
So? Why bother telling you all this? You have been given this
little opto-electronic engineering lesson simply because it will help
you to remember that you need to have your focus zone on
something that is not a single shade or colour for AF to operate
properly. This applies to all AF cameras, irrespective of brand or
model. So, if you encounter situations where the selected AF area
is not able to permit the camera to focus, check out the contrast
of the area and it's an odds-on bet that there will be little or no
contrast visible in the area.
The only deviation from this is offered by the use of a dedicated
flash unit such as the FL36R and the FL50R which will project a
pattern of light bars onto your subject (within certain rang limits
of course) in cases where no phases are detected (total darkness,
monotone surfaces etc.). This pattern is usually sufficiently
contrasty to allow the phase detection to take place.
Contrast measurement is achieved by measuring contrast
across a given point (usually midway between the sides of the
focus zone) by means of two sensors. The theory is that, at the
point of sharpest focus, the contrast graph as measured by the
two sensors will have the sharpest "edge" to any intensity
changes.
If, on the other hand, the lens is not focused sharply then the
"drop" will be more of an S curve than an L shape. The system
67