|
Claim 3 Full frame sensors provide
better image quality
The theoretical pixel-density advantage of crop cameras discussed
in Part 1 should have
given a victory to the crop-sensor cameras. Surprisingly, that
didn't happen, so let's see if the claim that larger-format sensors
provide better image quality is true.
A similar experiment was set up to that in Part 1, and again I
enlisted the services of a very co-operative chimpanzee. The lens
used was a Canon 70-200 f2.8 IS (with IS turned off) mounted onto
a sturdy Gitzo tripod .
However, this time, instead of shooting each camera from the same
distance to the subject, I attached each camera in turn and moved
the tripod backwards and forwards until the magnification/field
of view was the same for each camera. To judge this, I simply
placed two memory cards on the table, which enabled me to match
the frame width for each camera.
This methodology should put the full 16.7 megapixels of the 1Ds
mk II onto the subject - so should give this camera a huge advantage
- not only because it has a full frame sensor, but also because
it has 40% more pixels than the other cameras. The 1D mk III and
40D each have 10.1 Megapixel sensors and Canon's third generation
digic processor - but the sensors are of different sizes of course,
so it will be interesting to see if the the 1Dmk III with it's
larger sensor can beat the 40D in this test.
NB As in Part 1, I had to compensate for the different output
sizes from the 16.7 and the 10.1 megapixel cameras. To do this,
I opened the images in Photoshop at 100% (actual pixels) and then
used the generally-accepted practice of interpolating up the smaller
images and interpolating down the larger image to a mid point
value (actually 13.4 Mp). This method is not perfect, but each
camera is similarly slightly disadvantaged. In reality, it is
a small price to pay in order to make direct comparisons without
having to compensate for image size differences.
Starting image : Co-operative chimp. He stayed still the whole
time (unlike my cat that was my original chosen subject!).
100% crops (identically post-processed)
1D mk III 100% crop
40D 100% crop
1Ds mk II 100% crop
Just to show how good the 1Ds mk II is ....this is the 100%, sharpened
crop at original size with no interpolation.Look again at the
starting image of the chimp on the table to see what a tiny proportion
of the original image this represents.
CONCLUSIONS
As should be expected considering it's lofty original price tag,
the 16.7 Mp 1Ds mk II was the clear winner in this test. It shows
better resolution and the extra megapixels will enable much larger
prints to be made when compared to the 10.1 Mp cameras. However,
it is not possible to judge how much the full frame sensor is
contributing towards this performance and how much is simply due
to the extra 40% of Megapixels present. I must also confess to
having to search quite hard to find a subject that clearly demonstrated
the differences that I routinely see on my monitor. There is nothing
like feathers and fur (even if it is a bit synthetic) to show
up camera and lens differences.The quality differences also have
to survive the drastic downsizing that web-viewing necessitates.
The 1dmk III shows a small but definite and reproducible increase
in image quality over the 40D and this is in agreement with my
general feeling of having used these cameras for some time. However,
the differences only show up in the 100% crop. As the cameras
share Digic 3 processors, number of megapixels etc, it is reasonable
to assume that the increase in image quality is due to the larger
sensor benefits previously discussed. If you extrapolate from
the 1.3x crop sensor results then it is reasonable to expect that
a similar improvement again is present within the 1Ds II full
frame image.
All cameras tested produce stunning images, and it is only when
you start pixel-peeping at 100% crops that the differences become
apparent. The 1Ds mk II can clearly produce the best image quality
of the three, but at a price ! Whether you think that the extra
quality justifies the cost is a matter of personal opinion. I
bought my 1Dsmk II second hand after in was superceded by the
21Mp 1Ds mk III. The 1Ds mk III is essentially the same camera
as a 1D mk III other than it has the full frame sensor, so how
Canon can justify the massive price premium I don't know. At the
time of writing this in the UK, a new 1Ds III is £4700 compared
to £2376 for a 1D mk III - that's £2324 more, essentially
just for the better sensor !
The Canon 40D is currently only £498 and is therefore an
incredible bargain - as it can still compete with the other two
cameras on image quality. The 1Ds mk II feels like an old camera
by comparison, as it lacks the high-quality rear screen, liveview,
much improved menu system and requires a lot of complicated multiple
button-presses to operate it. The ol' gal can't half still deliver
on image quality though - it is a bit of a classic.
Finally I should give a big dollop of praise to the Canon 70-200
IS zoom lens (this is beginning to sound like the Oscars) as it
delivered incredible image quality throughout these test. It is
my favourite zoom lens as it can compete with most primes.
Claim 4 Full frame cameras have shallower depth
of field, so it is easier to blur the background
To put this to the test, I used my trusty chimpanzee assistant
again. The tests were set up as before with the 70-200 f2.8 IS
lens. I excluded the 1DIII from this test as the objective was
simply to determine whether reducing sensor size increased depth
of field. I had placed a newspaper at a fixed distance (two feet)
behind the chimp in order to help gauge depth of field.
The first test was conducted with the cameras set up at an identical
distance from the subject, with the identical lens set to the
same focal length (78mm) and aperture (f8). Due to the 1.6x crop
factor, the image from the 40D below appears much larger than
that from the full frame camera of course.
 40D
100% crop
1Ds mk II 100% crop
CONCLUSIONS
These are interesting results. Full frame cameras have a reputation
for having shallower depth of field - and this makes it easier
to get a smooth background in portraits for example. But this
is clearly not happening here - in fact it looks like just the
opposite is happening. What we are seeing is both the subject
and the background sizes being multiplied by the crop factor of
x1.6 in the case of the 40D. This makes the background lettering
appear more blurred. In the next image, I have interpolated the
40D image down by a factor of 1.6x to match the background size
(the text) of the 1Ds mk II image.
40D image reduced
Compare the depth of field again to the 1DsII image - now we can
see that there is actually no difference ! So let's say that again
- when shooting from the same distance,
using the same focal length lens and the same aperture there is
no difference in depth of field between a full frame and a crop
camera.
This is not actually the end of the story though, as another test
must be conducted, in which we will match the magnifications of
our chimpy subject. To achieve this, keeping everything else the
same, I moved the tripod/lens/1DsII closer to the subject until
the magnification was the same as the 40D image. I used a couple
of CF cards as crude markers of the width of the 40D field of
view, and then matched it with the 1Ds II in the viewfinder.
40D 100% crop
1Ds II image matched in magnification by moving the camera closer
to match the field of view of the 40D
CONCLUSIONS
When the full-frame camera is moved closer to the
subject to match the magnification of the crop camera,
with the same focal length and with the same aperture selected,
then we can finally see that the full frame camera
does show a marked decrease in depth of field. i.e.
blurring of the background. This can be a help or a hindrance,
depending on whether the requirement is to achieve greater or
lesser depth of field. Depth of field is at a premium in macro
photography, so a crop camera may be preferable to maximise depth
of field. However, when trying to separate a subject from it's
background, such as in portrait work, the full frame sensor's
decreased depth of field may be an asset. Horses for courses.
NOISE COMPARISONS
There are other comparisons that I have not carried out that would
be of interest if time were not an issue. Noise performance vs
ISO setting is certainly one. However, there are plenty of such
noise tests on other websites if you need this information:
Digital
outback
DP
review
< Back to Tips
and tutorials
< Back to Full frame vs crop part 1
|
