How AF Errors Decrease Effective Resolution of a DSLR/DC

http://ricehigh.blogspot.com/2006/12...effective.html

RichA, is that you?

You're article seems to misunderstand the focusing process. Adjusting the
focus changes the distance to the plane of best focus. The picture is still
in focus, just sometimes not at the distance that you want. Focus errors
are not limiting the resolving power unless you are photographing a flat,
perpendicular subject or a subject at infinity. Usually you just end up
slightly front or back focused where the plane of best focus is a little
ahead or behind the intended point. Take a picture of a wedge, or even a
tape measure held at an angle to see this.

What is really limiting the resolving power is the low-pass filter which
causes any sharp black to white transition to occur over approximately three
pixels rather than the one to two that would be ideal. However this
situation is far better than the moire effects that would occur without
this. Adding more megapixels does help this problem as it raises the
effective sampling rate and allows less low-pass filtering. Eventually the
lens would be the limiting factor, but it appears that we have not reached
this yet even with fairly cheap lenses as the resolving limits have
increased for the same lenses as the megapixel counts have increased.
Sharper lenses still help as the higher MTF near the limits improves edge
contrast.

That said, more accurate AF would be a great thing. Focus errors are very
annoying. Some camera bodies have a more accurate center point when using
lenses with apertures larger than f/2.8. Use this when you can.

I suspect some of the problem is with lenses that try to focus very fast.
There will probably be some overshoot as there is a limited amount of power
to accelerate the moving parts and stop them at the right spot. I suspect
this is why my Canon ef-s 60mm f/2.8 macro USM focuses so slowly, to try to
avoid this problem. Canon teleconverters also slow the focusing speed of
the attached lens while third party TC's do not which is why lenses
sometimes oscillate with third party TC's but not with the Canon ones.

The lens seems play a significant role in AF accuracy. I sometimes get
focus distance errors with the EF 50mm f/1.4 lens but almost never with the
f/1.8. Very strange. I would expect the opposite to be true as focusing is
done wide open, so the band of acceptable focus is smaller at f/1.4 compared
to f/1.8 and more light is available at the same focal length so it should
be easier to determine focus distance. Perhaps it is due to overshoot or
play in the drive train. The 50mm f/1.4 has an ultrasonic motor with
full-time manual focusing, but it is not a ring USM motor, but rather a USM
micromotor. Maybe this unusual feature causes play or overshoot?


"RiceHigh" wrote in message
ups.com...
http://ricehigh.blogspot.com/2006/12...effective.html

Not really. What you are talking is all about the amount of
out-focusing or simply the DoF (which is defined based on a certain
"acceptable" CoC) which the size of CoC can be calculated, for a
certain AF number and a certain aperture number and at a particular
focal length and at a particular object distance.

In case of an *obvious* AF error that can be seen, the CoC is actually
NOT accepted by the user, thus that's the ONLY problem. And the ONLY
cause of this problem is focusing error. Fo example, even if the
photographer shoot an infinity scene with all objects at infinity, a
serious focusing error can cause the resolution to drop significantly
owing to the large CoC.

Also, when I talk about out-focusing errors, of course I refer to the
object that is focused, and not other things in the same picture frame.

RiceHigh
http://ricehigh.blogspot.com/

default ¼g¹D¡G

You're article seems to misunderstand the focusing process. Adjusting the
focus changes the distance to the plane of best focus. The picture is still
in focus, just sometimes not at the distance that you want. Focus errors
are not limiting the resolving power unless you are photographing a flat,
perpendicular subject or a subject at infinity. Usually you just end up
slightly front or back focused where the plane of best focus is a little
ahead or behind the intended point. Take a picture of a wedge, or even a
tape measure held at an angle to see this.

What is really limiting the resolving power is the low-pass filter which
causes any sharp black to white transition to occur over approximately three
pixels rather than the one to two that would be ideal. However this
situation is far better than the moire effects that would occur without
this. Adding more megapixels does help this problem as it raises the
effective sampling rate and allows less low-pass filtering. Eventually the
lens would be the limiting factor, but it appears that we have not reached
this yet even with fairly cheap lenses as the resolving limits have
increased for the same lenses as the megapixel counts have increased.
Sharper lenses still help as the higher MTF near the limits improves edge
contrast.

That said, more accurate AF would be a great thing. Focus errors are very
annoying. Some camera bodies have a more accurate center point when using
lenses with apertures larger than f/2.8. Use this when you can.

I suspect some of the problem is with lenses that try to focus very fast.
There will probably be some overshoot as there is a limited amount of power
to accelerate the moving parts and stop them at the right spot. I suspect
this is why my Canon ef-s 60mm f/2.8 macro USM focuses so slowly, to try to
avoid this problem. Canon teleconverters also slow the focusing speed of
the attached lens while third party TC's do not which is why lenses
sometimes oscillate with third party TC's but not with the Canon ones.

The lens seems play a significant role in AF accuracy. I sometimes get
focus distance errors with the EF 50mm f/1.4 lens but almost never with the
f/1.8. Very strange. I would expect the opposite to be true as focusing is
done wide open, so the band of acceptable focus is smaller at f/1.4 compared
to f/1.8 and more light is available at the same focal length so it should
be easier to determine focus distance. Perhaps it is due to overshoot or
play in the drive train. The 50mm f/1.4 has an ultrasonic motor with
full-time manual focusing, but it is not a ring USM motor, but rather a USM
micromotor. Maybe this unusual feature causes play or overshoot?


"RiceHigh" wrote in message
ups.com...
http://ricehigh.blogspot.com/2006/12...effective.html