Rick Newton responded to me privately regarding my autofocus questions.  He
has asked that I forward those messages to SPORRS.  Here they are, at his
request.

John C. Lucas

Date: Thu, 25 Sep 1997 04:20:58 -0400
From: "Rick S. Newton" <[EMAIL PROTECTED]>
Subject: SPORRS: SPORRS revisited
Sender: "Rick S. Newton" <[EMAIL PROTECTED]>
To: John Lucas <[EMAIL PROTECTED]>
Content-Disposition: inline
X-MIME-Autoconverted: from quoted-printable to 8bit by
internet.roadrunner.com id CAA10629

Hi John,

I use an EOS-1n and a EOS-1 as my bodies.  Although lately I have been
using manual focus lately (I have been using a teleconverter that causes
unreliable autofocusing), I have used autofocus quite extensively, with
varying results.  

The EOS system uses a phase control something or other that compares the
contrast between two separate images to calculate the autofocus. Also in
the 1n, of the 5 focus points, only the center is cross biased.  In other
words, the camera can recognize vertical and horizontal lines in the center
focus spot, but only horizontal (or is it vertical, it slips my mind right
now) for the other four. First, the in-your-face headlights have a
tendency, because of their extreme brightness, to flatten the image, or
lose contrast per se.  The 1n then begins to fight this, causing eratic
focus behavior. With some of the bigger lenses, you can minimize this to an
extent by limiting the focus range, which isn't a bad idea anyhow, as there
is no need to focus down to 5 feet when shooting a train at 60 mph.  Now,
two different types of continuous autofocus are available, although you
actually only select one, the switch between the two is automatic.  First
is the normal continuous mode. Any focus spot on the camera will refocus
the lens when the object changes distance from the lens. However, with a 60
mph train barrelling at you, when you release the shutter, there is a
delay, about 6 milliseconds if I remember correctly, as that info is
transferred to the shutter. Then there is mirror travel time before the
shutter releases.  This is enough time to cause the object to become
out-of-focus.  Then there is predictive continuous focus.  This mode must
be started by aligning the object in the center focus spot. Once the camera
has been able to determine motion, the camera will track the object across
all focus spots, and will "predict" the correct focus of the object at the
time the shutter releases.  Unfortunately, with the EOS system, the object
must be travellin at a constant speed. Any change in speed, and you must
again center the object in the middle spot.

Here is the trick.  Most people will tighten their tripods down rigid, so
the camera may not move. For the in-your-face shot, tighten your ball head
down to where you can get just enough movement to follow the train a
little.  When the train comes into view, make sure you get that center spot
on something on the nose of the train, (handrails, logo, bell, etc), and
keep the spot on that object briefly to allow the camera to "lock-on". The
camera should then track the train across the other focus spots, provided
they don't get hammered by the headlights. Moving your composition slightly
as the train moves will avoid this problem.  I know you are going to say
that you don't want to change your composition, but, with autofocus, it is
a necessary evil. You may also say that you may not want the train centered
in your composition, but once the camera can lock on, it will follow the
train across the spots With a little practice, you can get it so you can
make such minor changes, you will never know the difference.

As far as the train not being covered by a focus spot problem, there are
two solutions. One, is very similar to the above. Leave your ballhead
slightly loose, and slightly move the camera till the train is in the dots,
get a lock, and AF lock the focus, then return to your precomposed
composition.  If the train is small enough that is does not cover one of
the spots, you may focus on anyhing near it, due to the lens at or near
infinity already.  The other way, is to simply prefocus  the camera, and
apply the AF lock.

Autofocus, unfortunately is still not a perfect science. It can easily be
fooled, although the manufacturer's are making it much better.  But with a
little practice, you r results can be fantastic. There is definitley a
technique involved, and when you find the niche, it becomes relatively easy
to great shots a vast majority of the time.  I look at it as no worse than
the many times I miss with the manual focus, although it usually is very
minor in nature.

I am going out on a slight limb, but I still think the EOS system is by far
the best. I wouldn't give it up for anything.  It takes a little getting
used to, but it is well worth it.

Hopefully , what I have said here helps. If not, Oh well, you can remove
the shoe from my mouth anytime now.

Sincerely,

Rick Newton
Die Hard Canon EOS shooter

Date: Thu, 25 Sep 1997 11:22:15 -0400
From: "Rick S. Newton" <[EMAIL PROTECTED]>
Subject: Re: SPORRS: SPORRS revisited
Sender: "Rick S. Newton" <[EMAIL PROTECTED]>
To: John Lucas <[EMAIL PROTECTED]>
Content-Disposition: inline
X-MIME-Autoconverted: from quoted-printable to 8bit by
internet.roadrunner.com id JAA24335

Hi John,

Glad some of my experiences  (not always good) have been helpful.

First on the metering issue, for the most part, you are correct, the
headlights and ditchlights throw AE off, sometimes as much as 2 or 3 stops,
which is totally unacceptable, especially  with AE, because with my
experience, AE exposures are already about a 2/3 of a full stop brighter,
usually due to lighter colored reflective subjects. There is a trick here
also, one that the EOS-1n, and actually I believe any of the EOS cameras,
and probably most of manufacturer's as well.  That being the one and only
AE lock. There is a very easy technique to this, simply, as the train
approaches, take a meter reading before the headlights are close enough to
affect the meter, lock the exposure.  If you are like me, and waste mega
film zooming back and firing as the train fills my viewfinder (by the way,
I have for the most part abandoned the zooming method), you would notice
that you exposure changes as the train size and reflective characteristics
change throughout the sequence. AE lock helps prevent this. Alas, unless
you suffer the horrible Cohen/Gilley phenomenom, where as the train
approaches, mere moments before the shutter is released, a large water
vapor filter obscures the sun, and not only is the shot hosed, but the
exposure is worthless anyway. Of course this happens when doing incident
light metering as well, unless you are superhuman and can take incident
meter readings, (and use them), as the train is blasting through your
viewfinder. :-) (sorry Dave and Jim).  Of course, also using spot metering
as well could provide a solution as well, provided the center of you image
is what you want to be metering.

Secondly, don't feel too bad about the lens situation, I only have 3 as
well.  28-70, 80-200, and a standard 50. Unfortunately, my 80-200 is the
older version, with the rear element flush with the camera mount.  I am
forced to use a 3rd part converter 2X for my lens, I cannot use the Canon,
which has the front element protrusion.  If you have the 70-200, you do not
have this problem.  However, with the Canon 1.4X converter (as well as the
2X) I understand autofocusing is good, however, it IS slower.  This may
account for part of your problem.  Autofocus cameras typically become
somewhat less "perfect" when slower lenses are used, because of the
perceived decrease in contrast that autofocus sensors use and that
limitation is about f5.6. Manual focus is required beyond that point. I am
using a 2X converter on an f2.8 lens, thus an effective aperture of f5.6,
thus slow and somewhat unreliable response, thus manual focus is preferred.
  Using a 1.4X brings it to f4.0, pretty acceptable as far as the camera is
concerned, but definitely slower, and a little less reliable.  As you learn
the system(s),  you become accustomed what the system can and cannot
accomplish (which you seem to have a pretty good handle on).  Even at f5.6,
the autofocus can work fairly well for me on a train moving at or near a
perpendicular motion in relation to my lens, but I know if that damn thing
is barreling nearly head on, ibetter be manually focusing.  Of course
usually, these things are decided by your composition long before the train
ever arrives.

Here is another example, when I know I am goin to use the teleconverter,
and I have to manually focus, I use the older EOS-1, because the focus
spots are not in the way.  When not, the 1n is usually the camera of
choice.  I typically keep the same film in both cameras, so my incident
metering is correct when I switch bodies, or are using both bodies
simultaneously.  I do incident metering as well, except when I am in shade,
and the time or distance is too great to walk into the sun for a reading. 
BTW, I usually only use the 28-70, and the 80-200 plus the 2X tele. I find
the 28-400mm range is generally good enough for my photography. Now if I
ever get enough money, the 17-35 and a 300 2.8 could be on my horizon
though.


Rick




--> SPORRS: Serious Photographers of Railroad Related Subjects


Reply via email to