|
Canon 1DmkIV - Early impressions, 12th March 2010
|
Canon 1DmkIV 16 mp 1.3x crop Digital
slr |
I have been testing the new Canon 1DmkIV this month. The main
issues I was interested in were:
1) How does it focus ?
My two "post-submirror fix" Canon 1DmkIII's have focused
very well in AI servo and I have many great action images from
them. I never experienced the kind of issues that some users unfortunately
report, but I am happy to state that the mark IV autofocus is
very clearly superior from the outset. Initial acquisition is
still very snappy whilst tracking is notably better and the number
of in-focus images in a burst is significantly improved too.
My first quick AF test of a new camera is always to tripod-mount
it and ensure that after initial focus acquisition, it will focus
stably on a static subject in AI servo without continuously attempting
to refocus (chatter). My new mkIV passed this test fine, as did
both my previous mkIII's. You can't expect your camera to focus
on something moving if it can't stably focus on a static target.
I had read that some people find that the mkIV will not focus
as well as the mkIII in low light, so I tried focusing on some
objects in my living room in the evening in very low light. I
found that again, focus seemed absolutely fine, and appeared no
better or worse than the mkIII as far as I could tell.
I then had to carry out lens micro-focus adjustments to ensure
I was not experiencing any front or back focus issues. I constructed
a paper target mounted on a board at 45 degrees. The lens under
test was mounted on the camera/tripod and using a cable release,
I focused at the lens's widest aperture - on the black line at
the centre of the test chart. I usually place the tripod at a
distance from the target that seems relevant to the lens focal
length, that is, where I can read the chart numbers and determine
if they are in or out of focus. Canon suggests using a vertical
target and testing at 50x focal length. I find that this is OK
as a final check of focus but this method gives no indication
which way to carry out the adjustments and the distance seems
excessive to me.
I take a quick series of photos at various plus and minus micro
adjustment settings, then view them on the back of the camera
on the new super-sharp screen and determine whether the camera
appears to be front or back focusing. I make the camera refocus
each time by defocusing it first.
If I think that the camera is front focusing for example, I take
a series of images in triplicate from zero backwards and then
view them on my computer screen at 100%. The appropriate camera
lens micro adjustment is dialed-in to the camera so that the numbers
in front and behind the focus line on the chart are equally in
focus.
The chart that I use is available here : Tim
Jackson lens adjustment chart
I like the fact that the Liveview now offers autofocus (unlike
the mkIII which is manual focus only). In Quick Mode the mirror
has to flip down, the camera focuses and the mirror flips back
up. This makes a bit of a clack but it is very quick. This ability
now permits you to focus on a fixed target, zoom in to 10x and
see if you can improve on the camera AF by focusing yourself manually.
When you can't better what the camera can do yourself you know
the adjustment is correct.
Finally, there is another method of conducting lens micro focus
adjustments with the camera tethered to a computer/laptop. It
is a clever, accurate and novel method : Lens
micro focus adjustment tricks. All you need is the USB lead
and the Digital solutions disc that ship with the camera. I found
that once I had set up my lenses using the 45 degree target method
I got very good correlation when I rechecked using the tethered
method. I found that my 500mm f4 required no adjustment but some
other lenses were front-focussing and needed adjusting backwards
to compensate.
I was now satisfied that my camera was set up properly and I
then moved onto taking pictures of my cat as she ran around the
lawn. I used ISO 2500 to get a high shutterspeed (1/3200sec at
f6.3) on the 70-200 f2.8 IS lens. I selected expanded centre pont
focusing in custom function CFIII- 8 and focussed on the cat's
head as best I could as she dashed around.
100% crop (ISO 2500)
Every one of the 22 images in the burst I took were in focus.
The 1DmkIII could never achieve this performance, it would nearly
always threw up several out-of focus images in the sequence.
Other tests I conducted included my cats running towards me (below)
, some tests of a jogger running towards me using the lens wide
open this time at f2.8 - and the hit rate was near 100% when I
got it right ! This is similar to the performance of the Nikon
D3 I hired some months back.
Although I could not test them side-by-side, I feel that the 1DmkIV
now equals the D3 in tracking and has the added benefit of quicker
initial focus acquisition. I set the appropriate focus limiter
setting on the lens and then try to pre-focus at a similar distance
to my subject so the camera does not have to hunt backwards and
forwards to "see" something to lock onto. Once I attain
focus I keep the shutter button half-depressed and take short
bursts of images when the time seems right. If I lose focus, I
release the shutter button and repress it when I feel I am on
target again.
Finally I moved onto some wildlife subjects - some herons flying
around and building nests - using my 500mm f4 IS lens mounted
on a Wimberly mkII head/gitzo tripod.
I found that my keeper rate was certainly much higher than with
the 1D mkIII, which had a tendency to throw up a few out-of focus
shots in most continuous image bursts.The mkIV images that were
sharp were usually "tack" sharp - not just "pretty"
sharp as they sometimes were with the mkIII. When sniping quickly
at ducks and gulls I was impressed at how many were in focus.
If I could keep the AF pooint on the target, I felt that the camera
did not try to jump to the background to the same degree as the
mkIII. There is still a very high degree of skill required from
the user and I can usually say that if the focus missed that it
was usually my fault - so no excuses now! I experimented with
centre point only, expanded centre point and all 45 points and
I would say that my findings are in line with Les
Zigurski's excellent article .
I can't understand how some testers feel confident enough to analyse
and report back on AF capability with great certainty, as it is
anything but an exact science. I think that the best that can
be done is to build up impressions over time with continual use
and practice. I never found that fiddling with the custom functions
made a huge amount of difference on my mkIII's performance and
I would say the same about the mkIV but it is still early days.
However I find I am getting good success with pretty much "out
of the box" settings but I change from single point to single
point with expansion or all 45 points active according to background
complexity and how erratically the subject is moving.
2) What is the image quality like compared
to the 1DmkIII and also my reference 1DsmkII full frame camera
?
The 1DMkIII is capable of producing top flight image quality.
With the 60% extra pixels I was hoping that the mkIV would produce
similar or even better image quality but with higher resolution
than the mkIII. In a nutshell that is pretty much what I see in
the mkIV's images. The colours coming out of the camera seem a
little more vibrant and hold a lot of detail. You can see this
in the highlights of this swan below. I shot RAW and I used the
Photoshop CS4 shadow/highlight tool a little to maximise detail..
|
| Mute swan - even at this
tiny web size, the amount of detail in the silky
head feathers is very apparent. |
|
| 100% crop of the above
image |
Next I carried out some side-by-side comparisons of the 1DmkIV
against the 1DmkIII and the 1DsII with a 100mm f2.8 macro lens
fitted in turn to each camera.
In this test, I balanced a feather on the nose of my trusty assistant
which I figured would be a good test of resolution in the field.
I framed the images identically by moving the camera/tripod. This
means that the full frame camera was 1.3x closer than the crop
cameras. This mirrors a situation where the subject will allow
you to get as close as you want and effectively gives the full
frame camera its best chance to excel.
 |
Starting image |
|
| 1DsII 100% crop - The image
is larger because it has more pixels |
The images from the three cameras have a family
resemblance. The 1DmkIII image is smaller and has lower resolution
as it has 10Mp compared to the 16Mp approx of the other two cameras.
It consequently appears a little sharper at 100% ( but this differences
disappear if the other images are downsized to mkIII size).
I was surprised at how well the new 1DmkIV with its 1.3x crop
sensor holds up against my reference - the full frame 1DsmkII.
There is virtually no difference at 100% in this test.
The next test was performed at the same tripod to subject
distance in each case. This is the situation where a subject
may not allow you to get any closer (Focal length limited scenario).
These are 100% crops of an old house taken with a 17-40mm f4 lens
set at 25mm.
|
1DmkIII - 10Mp on a 1.3
crop sensor |
|
| 1DmkIV - 16MP on a 1.3x
sensor, larger than mkIII because it has 60% more Mp on
the same size sensor (higher pixel density). |
|
1DsmkII - 16.7 Mp on a full
frame sensor. The image is smaller than mkIV as a) The full
frame camera has a wider field of view and b) It has has
same number of pixels as the mkIV but on a larger sensor
- so pixel density is lower. It has less resolution than
the mkIV at this distance, but still resolves a just little
more than the mkIII.
|
These images have been sharpened equally (300%,
0.3,0) and the mark III has the lowest resolution but is the sharpest
at this pixel level. Larger images usually require firmer sharpening.
I thought that the mkIV images were a little soft at first before
I realised that they require (and take well) a firmer hand with
sharpening.
|
| 1DmkIV image sharpened
a little more.
Now matches mkIII image for sharpness but with more resolution.
The benefits of the high pixel density for telephoto work
are now very apparent for situations when you can't get
any closer to your subject. |
To give you a feel of how the high pixel density
can be used to advantage, here is a picture of a hare which was
a little small in the frame as I couldn't get any closer. If you
look at the 100% crop beneath it you can see how hard the image
can be cropped and still look sharp, but just as importantly -
natural. The full frame 1DsmkII has the same number of pixels
and great pixel quality as the 1DmkIV but the image would have
to be cropped 1.3x harder for the same framing as the field of
view is wider. There will be less pixels on target as the pixel
density is lower - so the 1DmkIV has a telephoto advantage.
|
Brown hare taken on a very dull day at ISO 800 with 500mm
f4 IS plus 1DMkIV, beanbag, car as hide.
|
|
| 100% crop of above image. Very good detail at
this magnification would permit me to crop the image to pretty
well any size I want for printing. |
3) What is the ISO performance like compared to the 1DmkIII
?
This is a low light test of ISO carried out indoors on an old
collection of butterflies (with the glass removed). The lens used
was a 100mm f2.8 and I used a tripod/cable release and mirror
lock up as the shutterspeeds were very low. The focal point was
the head of the black and green birdwing butterfly at the case
centre. The butterfly scales provide a good indicator of detail
and the black of the butterfly and the white background show up
noise well.
Starting image :
100 % crops (using ACR 5.6 as RAW converter) :
|
|
1d4 100 iso |
1d3 100 iso |
|
|
1d4 200 iso |
1d3 200 iso |
|
|
1d4 400 iso |
1d3 400 iso |
|
|
1d4 800 iso |
1d3 800 iso |
|
|
1d4 1600 iso |
1d3 1600 iso |
 |
|
1d4 3200 iso |
1d3 3200 iso |
 |
|
1d4 6400 iso |
1d3 6400 iso |
 |
|
1d4 12800 iso |
1d4 6400 reduced to 1dIII size |
 |
|
1d4 25600 iso |
1d4 reduced to 1dIII size + Neat image |
 |
|
1d4 51200 iso |
1dmkIV 3200 iso - DPP |
|
|
Nikon D3 iso 3200 |
Nikon D3 iso 6400 |
Conclusions on noise
There is a problem when viewing images at 100% size when they
are from cameras of different Mp counts. The larger output camera
should show more detail but could appear a little softer and with
more noise as you are looking at it much bigger on the screen.
Both the 1DIII and 1DIV when correctly exposed look very impressive
up to 1600 iso. Both cameras retain very good detail up to 6400
iso, with fairly equivalent noise when the size difference is
taken into account. I reduced the 1dmkIV image to 1dmkIII size
and the images are pretty much identical. I also
applied Neat image noise reduction to the reduced 1dmkIV image
and it does look very usable. This was in line with my expected
level of performance from the 1DmkIV, but I was really surprised
that the 1DmkIII came out as well as this too.
The 1dmkIV can go on to be used at iso settings of 12800, 25600
and 51200. As far as I am concerned these settings are only useful
for small web-sized viewing and are just marketing gimmicks. The
two highest settings are just plain daft in my opinion.
I have a couple of 100% crops at 3200 and 6400 iso taken when
I hired a Nikon D3. I have included these above in the comparisons.
Although they are of a different subject taken on a different
day, they give a good idea of the Nikon D3's noise capability.
They look pretty similar to the two Canons at the same iso - I
was surprised that the 1DmkIII was so similar to the D3. The latest
Nikon D3s is said to have 1 stop better noise than the D3. So
Canon has done well to retain the same noise performance as the
1DmkIII despite 60% more pixels on the same sized 1.3x sensor,
but to expect it better on noise and be competition for the D3s
in this respect is wishful thinking.
Finally I wondered how a 3200 iso image from the 1DmkIV would
look when converted with the Canon DPP software. It does show
a little less luminance noise (granularity) than the ACR5.6 conversion
but worse chrominance (colour ) noise. So I will continue to use
ACR as my preferred converter.
In summary then, Canon have retained the good noise performance
of the mkIII but I am disappointed to report that I can't demonstrate
evidence of any actual improvement in my tests although this is
hard to establish conclusively as the filesizes are significantly
different.
Plus points of the 1DmkIV
I am very pleased with the autofocus ability of the MkIV in both
one shot and AI servo. It is not perfect - nothing is, but it
is a definite improvement on the mkIII and appears to be at least
a match for the Nikon D3 I tried some months back but I would
like to be able to compare the two cameras side-by-side to come
to a positive conclusion.
Great resolution and general image quality from the 16Mp sensor
is very satisfying. The files appear to look very natural, take
cropping and firm sharpening well and lack any on-board noise-reduction
artifacts that I have observed on othe rcameras such as the Canon
40D and 50D 1.6x crop cameras. I was expecting that ultimate image
quality would remain with the full frame 1DsmkII, but I was surprised
and pleased to discover that they were remarkably similar. I suspect
that only the 21 MP Canon 1DsmkIII or 5DmkII or 24Mp Nikon D3x
could beat it.
High pixel density and good per-pixel quality is excellent for
working with more distant subjects and should give the Canon 1DmkIV
a big advantage over the 12 Mp full frame Nikon D3s when shooting
side-by-side in a focal length limited scenario. This is often
the case with birds.
The high resolution of the rear screen is very nice - it is now
possible to judge image sharpness at last. The larger RAW buffer
is great, and I like the fact that different focal points may
be selected in the landscape and portrait formats. I have the
portrait orientation set to be over a point that usually coincides
with a bird's eye. I am pleased that 45 points are user-selectable
again rather than the 19 of the mkIII . Also the crazy method
for selecting AF points has been abandoned for the far better
earlier system found in the 1DmkII.
The mkIV has HD video - but I can't comment on that as I have
yet to find a use for it. Autofocus in Liveview is a welcome addition
though for macro and hide work.
Minus points
Although I did not have the Nikon D3s to compare the 1DmkIV against,
all indications (from other Internet reviews) are that it has
about 1 stop better noise performance than the D3 and so it will
beat the 1DmkIV by about the same amount in this respect.
The high pixel density of the 1DmkIV has two negative implications
- it puts great demands on lens quality and and I suspect that
the camera is more susceptible to camera shake/motion blur. I
find that I have to raise my shutterspeed a little to ensure getting
a sharp image on the mkIV. I'm not yet certain whether this is
because the mkIV has a pretty harsh mirror slap that sets up internal
vabration (the shutter certainly makes a fair old clack) or whether
this is simply because at 100% on the computer screen, the image
is a lot larger than say a 1DmkIII and you are looking that much
closer at the imperfections. I have explained this issue in another
tutorial called - Do
high megapixel cameras produce softer images ?
Although I did not carry out any scientific measurements, the
battery life appears to be a little worse than the mkIII even
without using video, this is presumably because the processors
are working harder.
Conclusion
The 1DmkIII is a great camera that I would recommend
it to anyone, but the mkIV is altogether much improved - it is
better in many areas, notably AF performance and resolution. Although
still impressive, I was unable to demonstrate that Canon have
improved on the noise performance as claimed and I am sure that
its rival the Nikon D3s would better it by a stop or two. Also
although the mkIV is capable of great image quality, I find that
the camera puts greater demands on the photographer's technique
and also lens quality if the improved performance is to be realised.
The mark IV is certainly an evolution rather than a revolution
but the improvements all add up to make it an even better wildlife
camera than its predecessor - with the bonus of HD video if required.
I feel that Canon has got the balance of filesize/pixel density/image
quality just about spot-on for my kind of application so I would
strongly recommend it.
Update 22nd March
I was able to carry out a side-by side comparison test of the
21Mp Canon 5DmkII against the 16Mp 1DmkIV.
I used the case of butterflies indoors in low light again. The
lens used was a Canon 180mm f3.5 macro lens, Other equipment included
a Gitzo tripod/Markins ballhead, and as shutterspeeds wer low,
I used mirror lockup.The tripod was moved closer with the 5DII
to get the same framing as the 1DmkIV.
Test 1 - Low ISO image quality
|
1DmkIV 100% crop - 100 ISO |
|
5DmkII 100% crop - 100 ISO |
Test 1 conclusion.
The 5DmkII image is larger because it has 12.5% more pixels than
the 1DmkIV. Pixel peeping at 100% shows that the 5DII has a little
extra resolution and apparent sharpness. The image quality (IQ)
on both cameras is excellent but the full frame camera with the
extra pixels wins out as would be expected. What is pleasing is
that the mkIV comes so close to Canon's best IQ camera despite
the smaller sensor.
Test 2 ISO comparison
|
|
1DmkIV 100 ISO |
5DmkII 100 ISO |
|
|
1DmkIV 200 ISO |
5DmkII 200 ISO |
|
|
1DmkIV 400 ISO |
5DmkII 400 ISO |
|
|
1DmkIV 800 ISO |
5DmkII 800 ISO |
|
|
1DmkIV 1600 ISO |
5DmkII 1600 ISO |
|
|
1DmkIV 3200 ISO |
5DmkII 3200 ISO |
|
Image not available |
1DmkIV 6400 ISO |
5DmkII 6400 ISO |
Test 2 conclusion
The 5DmkII shows slightly less chrominance (colour) and luminance
(granular) noise at each ISO setting - equivalent to approximately
1 f stop. Bearing in mind this is extreme pixel-peeping, the results
are very impressive on both cameras. Unfortunately I appeared
to mislay the 5DmkII 6400 ISO file somehow - doh ! but the results
are still pretty conclusive - both are really usable at 3200 ISO
and 6400 ISO at a pinch. I did not explore ISO settings beyond
6400 ISO as I am usually interested in high quality images rather
than obtaining an image at any cost.
Overall conclusion
The 5DmkII with 12.5% extra pixels on its full frame sensor has
about 12.5% higher resolution and about 1 stop lower high ISO
noise than the 1DmkIV. This was to be expected, but the differences
are pretty minimal even on these huge 100% crops. In the real
world, the differences are slight, but the 1DmkIV has other strengths
- a much higher 10 frames per second shooting rate and much higher
spec autofocus being the main benefits. As such, the 1DMkIV is
a brilliant all-rounder and the best wildlife camera on the market
in my opinion at the moment.
|
