Background
I was carrying out some tests on a couple of new lenses on a dull
day and my shutter speed was fairly low at around 1/80th sec.
Despite focussing carefully and using a good tripod and wireless
remote shutter release I was disappointed to find that most of
my shots were very soft. I had not used mirror lock- up as I thought
that my shutterspeeds were sufficiently far out of the 1/10-1/20
sec danger zone that is often cited. At these shutterspeeds vibrations
resulting from the camera's mirror rising and falling are known
to produce vibration that can cause blurring of detail. In order
to see what was going in my case, I decided to set up some tests
to investigate a bit further.
The first set of tests were done with the Canon 5D mkII camera
and 100mm f2.8L IS macro with tripod collar fitted. I wanted to
see how much mirror slap affected the images as shutterspeed was
progressively reduced. I figured that as the 5DmkII has a fairly
soft-sounding shutter it would not be causing too much
vibration. Also a 100mm lens is fairly short and would probably
not amplify vibration to the extent that a longer telephoto lens
would. So I started with this combination.
Everything was done identically with tripod, cable release, IS
off, camera in One shot (centre point only focussed on the centre
of the banknote).
I allowed 5 secs between each shot for the camera vibrations to
settle down, lens was micro- calibrated and identical processing.
ISO used was 400.In the first test I let the camera focus each
time (a slight mistake as it has introduced a little variation
between shots) but in the second test with the 1DmkIV and 70-200
f2.8 L IS mkII, I switched to manual focus to prevent refocussing
errors creeping in.
Test 1 Starting image
The target was a banknote with an engraving of a mandolin
on it. I focussed on the centre of the frame - on the curved portion
of the instrument.
100% crops of the banknote.
Results
From test one some interesting things emerged:
Firstly, before I get onto mirror slap, there is a nice demonstration
of the effects of diffraction here as it can be seen to progressively
creep in as the aperture size is reduced. Just take a look down
the right hand column of the mirror lockup series, you can see
that the first hint of softening due to diffraction creeps in
at f11 and gets increasingly worse as the aperture gets smaller.
F16 is still fine - and is the aperture usually recommended for
macro work. I would still use f22 if pushed and just increase
the sharpening a little.But f32 is decidedly soft. No surprises
there but this reinforces the need to be wary of using very small
apertures.
The non mirror-lockup series will also have this diffraction superimposed
upon the effects of any mirror slap/vibration. What surprised
me was that I could see the effects of mirror slap so much earlier
than I would have expected. Unfortunately the first non-MLU
shot at 1/320 sec is a bit soft - presumably due to a slight focusing
error as the next 1/160 shot is a bit sharper. However, if I was
generous and said that the 1/320 shot was probably still unaffected,
I can still see the beginnings of softness at 1/160 vs the MLU
version. This softness is present all the way down to 1/40 vs
the MLU version but at 1/20, 1/10th and 1/5th it gets quite a
bit worse.
This is the well-documented zone of shutterspeed where MLU is
usually recommended. By 1/5 to 0.4 sec the mirror slap has passed
again and the non-MLU version matches the MLU version which is
also soft due to diffraction.
Test 2.
This was performed exactly as before but with a longer lens,
the 70-200 f2.8L IS MkII.
This time I was shooting from a greater distance of course to
get a similar fiedl of view and being a physically longer lens,
I suspected it could be more prone to amplifying mirror slap.
I mounted a 1DmkIV camera onto the lens this time as I was wondering
if my 5DmkII could be faulty in some way. The 1DmkIV in common
with all 1 series cameras has a noisy and rattly sounding shutter
(presumably as it has to be more robust and can perform at a higher
10fps frame rate.) and I wouldn't have been surprised if it fared
even worse than the 5DII in terms of mirror slap.
Once focussed, I locked the Autofocus into manual which I hoped
would remove one more variable. I also increased the ISO to 1250
as I wanted to start at a higher initial shutterspeed than before.
Here are the results:
Results
These results appear a little more consistant than the first
set as there is no refocussing between shots. As expected, diffraction
was present from f16 onwards even in the MLU series - f32 being
very bad.
I think that you can see that mirror lock up has a signicantly
beneficial effect at 1/125 and below
and from 1/60th sec to 1/8 sec, mirror slap is very significant
without it and this explains why I was getting poor results befoer
conducting these tests.
Conclusions
I think that these shutterspeeds are much higher than you would
normally expect to see the effect of mirror slap and should consider
using MLU or Liveview shooting (when the mirror is already raised)
when shooting off a tripod with remote shutter release at 1/125th
sec or below - particualrly when using a 1 series camera.
I would also allow around 5 sec for vibrations to subside between
shots when shooting in this way. When shooting with the camera
tethered to a computer and viewing on the screen at 200% I could
clearly see oscilating movement of the image on the screen for
at least 3 secs after remotely closing the shutter with the computer.
The effect of mirror slap is definitely worse with the longer
lens/1DmkIV combination but the effect was still present in the
macro lens/5DII combination. I have not proven that one camera
body is worse than the other (although I suspect that the 1 series
is worse in this regard) - this would take more investigation,
but the overall message here is the same.
It should be said that this is a very stringent test as the cameras
are 21Mp and 16Mp respectively and the 100% image size is very
large when seen on the computer screen and we are looking very
critically at a very small part of the image. However, if
we want top quality images from our high-resolution cameras, we
must consider resorting to stricter vibration control when working
off of a tripod.
In the case of the 70-200 f2.8 L IS, results would probably have
been better handholding down to 1/30 sec thanks to the lens's
great 4-stop IS system and the photographer's hands damping the
camera vibrations.
Summary
The effects of mirror slap appear to be encountered at much faster
shutterspeeds than is usually suggested. The reason for this is
not revealed by these tests, but one can hypothesise as to whether
thismay be because later cameras are producing more vibrations
than previous models or whether high megapixel camera sensors
are responsible for revealing the effects much earlier than before.
When shooting on a tripod with a cable release, I would
now want to use MLU or shoot in Liveview mode or at least try
to damp the lens vibrations with a beanbag placed on top of the
lens/camera at shutterspeeds of 1/125th sec or less. In the past
I would not have resorted to this before I hit 1/10- 1/20th second
!
When using very long lenses such as the 500mm f4 off of a tripod
I will continue to shun using a cable or wireless release but
continue to recommend using "good long lens technique"
which is described in part
5 of my article on photographing wild birds.
|
