Canon EOS 1D MkIV
Canon announced the new 1D MkIV on the 20th October 2009. We all waited with great anticipation for this new body to arrive in December 2009. It didn’t, I got mine on the 11th January 2010. I was very excited to test the new body, especially after the debacle with the focusing of the 1D Mk III.
So, before I get to the results, and I must add preliminary results, as I have not had too much time to test the body, let’s look at what is new or better according to Canon. Once again, I am not going to comment on the video capabilities.
So, let us look at the tests.
Below are images photographed in the field, it usually tells a complete different story than the studio images. Here, you clearly will see how the noise stands out, and again we have only taken images from ISO 400 above and did not bother to go higher than ISO 6400. The thought here was that the MkIII could not cope with anything higher than ISO 3200 so we limited our field test to ISO 6400, in hindsight, a mistake and you will see later on why.The camera settings and process of the images were as follows:
Remember what Canon has done here to the MkIV, they have increased the amount of pixels from 10Mpixel to 16Mpixel making the individual pixel size smaller from 7.2m on the MkIII to 5.7m on the MkIV. This should increase digital noise, but as we have seen, the MkIV’s noise to me is the same as the MkIII’s noise at the same ISO and better with the additional pixels, is it better than Nikon’s D3S? I don’t know and I will have the opportunity of checking that within the next week or so, BUT, I need to add once again we are not comparing apples with apples as the D3S is a full frame sensor and only has 12MPixels. The only exception I have noticed is that the MkIV’s noise was slightly higher on ISO3200 than that of the 1DMkIII. Although Canon claim that the MkIV can photograph at an ISO 102400 the images are totally unusable, anything above ISO 25600 is not usable even if you apply the suggested noise reduction in DPP the image becomes soft and will maybe only be usable in very low-resolution reproduction and nothing more. See examples below:
After doing these tests you realize how good the noise is on this camera!!! Look at the ISO numbers we are talking about, ISO 12800 who would have thought that you can photograph a usable image at that ISO 12800 and not only usable but you can print it up to 15 x 10 inches and have a very good print. I will never suggest that you photograph at those ISO’s but, for sport photographers under floodlights ISO 6400 – ISO 10000 will give you a very good image. What is very important when you are going to photograph at these ISO’s is that your backgrounds needs to be nice and light, the moment the backgrounds go darker the noise becomes much more evident.
Now we are coming to the main problem that the MkIII had, inconsistent focus, even with static subjects and not good at all with moving subjects. We derived two tests here, one for static subjects and one for moving subjects.
Let's look at the static subject, and what we wanted to see is if the focus point focuses at the point where you aim the focus point and, how does the different assist focusing points influence the main focusing point. For the static test we printed a test chart from Norman Koren at http://www.normankoren.com. The chart is really to test lens resolution but we thought that this gives you a very nice chart with clear black and white lines that will create ample contrast to focus on. We used a Canon 400 f2.8L IS USM lens on a sturdy tripod with a Wimberley head.
We are not going to include all the test images, there is just too many, but you will get an idea of how the test was conducted and we will show the problem areas.
The above images show the use of all 45 focusing points, and you can see that the camera selected the centre focusing point to focus with. In all the other tests where we used the centre focusing point, see combinations below, the camera focused perfectly on the spot without any hesitation or hunting and was not distracted by any of the expanded focusing points (in both One Shot Focus and AI Servo Focus).
This sounds quite obvious, but remember what Canon says about using expanded focusing points and I quote, “If the manually-selected AF point cannot achieve focus, focus can be achieved with one of the adjacent (expanded) AF points”. This means that the manually selected focusing point must be used first to achieve focus and if the manually selected focusing point cannot achieve focus only then will the expanded focusing points search for focus. Out of experience with earlier models and with smaller bodies (entry level and semi-pro bodies) the camera will focus with the closest expanded focusing point ignoring the selected focusing point. In the case of the MkIV we did not get any such results with the centre point, so we thought, until we did a closer inspection on the images.
In the next set of tests we checked the outside focusing points. See thumbnails below:
Bottom AF Point
Top AF Point
Left AF Point
Right AF Point
Here the bottom and top focusing points worked perfectly (the first two images from the left). When it came to the left focusing point (third image from the left) we had poor results. The right focusing point (far right image) gave us mixed results. We will deal with them separately; when we used only the left focusing point without any expansion, the focusing hunted and it eventually achieved some sort of focus as you can see from the image below the focus was not accurate, the focus is in front of the target, and it did it with both focusing modes:
Left AF Point
When we added expansion to the left focusing point the result was as follow:
We experienced the same problems with the right focusing point but with less severity. When we used only the right focusing point without any expansion the camera hunted slightly and achieved focus. See images below, the first image is with One Shot Focus and you can see that the focused area is in front of the target. The second image is with AI Servo and the focused area is within the target area:
One Shot Focus
When we added expansion to the right focusing point the result was as follows:
So, with added expansion we did not get faster focus, the camera hunted slightly and then achieved focus with mixed results between One Shot Focus and AI Servo Focus. With the left focusing point, we never achieved focus on the targeted area. This made us wonder if there was a problem with the chart that the left and right focusing points struggled and with none of the others. So we moved the chart so that the right focusing point can focus on the area where we did the centre-focusing test, and then did the same with the right focusing point. See thumbnails below:
The result was that with both focusing points left and right, and with all three expanded options, instantaneous and accurate focus was achieved. So, we moved the chart so that we can focus with the centre focusing point on the original positions of the left and right focusing points. Will we achieve fast and accurate focus from the centre point? Well, here is the results see thumbnails below for the positioning of the centre focusing point, the left thumbnail shows the centre focusing point in the original left focusing points position and the thumbnail on the right shows the centre focusing point in the original right focusing point:
Let’s deal with the left image first, remember this is the one where the left focusing point was on originally and the left focusing point really struggled, not once did the left focusing point hit the target. So this is what the centre point did without any expansion:
The focus was achieved immediately but the same as the left focusing point in front of the target. When we added expansion to the centre focusing point the results was as follows:
Let’s look what the centre focusing point did when we put it on the original right focusing points spot. Centre point without expansion:
And the answer is spot on!! With expansion we got the following results:
The last test we then did was to use spot focus and we did again with the centre point, top, bottom, right and left focusing points, see thumbnails below:
All the focusing was achieved immediately with no hunting, and once again centre, top and bottom was spot on without any problem. The right focusing point was also within the target area but towards the bottom of the focusing point. The left focusing point had about 50 % of the target in the bottom part of the focusing point and the other 50% is below and outside the focusing point, but much better than any of the other tests above.
See the three images below; the left image is the left focusing point, centre image the centre focusing point and the right image the right focusing point:
The centre, top and bottom focusing points worked perfectly, but we suspect it is because the target areas were very easy to pick up without any distraction around the target area. The left and right focusing points struggled as well as the centre focusing point when we moved it to the more complex target areas and, we suspect, if we moved the top and bottom points to the same area we would have had the same inconsistent results. The obvious question now is why did this happen? Well one of Canon’s problems always was that focus will always be achieved on the closest subject, meaning if I use all 45 focusing points and focus on a subject with lots of contrast, focus will be achieved on the closest point, and it will even override the more sensitive centre point. That was the reason why I gave you that quote from Canon earlier. Now with the tests it looks like they sorted the problem, especially when we moved the centre point to the more complex area, the centre focusing point could not achieve focus, hunted and then used one of the expanded points, as what Canon’s statements says. BUT why, when we use the more complex area, is all the focusing achieved in front of the target area? The only explanation I have is two fold, one the old problem of focusing on the closest contrast area the focusing system can find and secondly the NEW LONGER FOCUSING HAIRS. The combination of the above two reasons makes a lot of sense, especially if you look at what happened when we used the Spot Focus option, remember the Spot Focus reduces the size of the focusing hair. Have a look at the diagram below:
See also below a side-by-side comparison of the left focusing point normal without any expansion (on the left) vs. the left focusing point with Spot AF activated (on the right).
Above you can see that both focusing points achieved focus but the focus is in front of the targeted area, spot focus (right image) less than the normal mode. For those of us that do wildlife photography (long focal length lenses close to your subject) and for those photographers that work with very shallow DOF take note of this problem. When you think that there is a chance that the focus point can be influenced by a nearby distraction use the Spot Focus. I also suggest that you select your focus points manually and only use surround expansion (CFn III 8-2) when you really need too and if your subject is parallel to you. I feel more comfortable with the left/right expansion (CFn III 8-1) than the surround. Stay away from the 45-point expansion (CFn III 8-3) as it focuses on anything it can find. See two field images below shot right after each other, although the centre focusing point is on the Hippo the focus first focused on the water next to the Hippo and the next image on the grass bank at the back.
We need to do many more tests on the focus expansion in the field and also remember these tests were done with the old firmware 1.04.
For the second test, a moving subject, to test the tracking ability of the focusing system, we came up with a test to minimize “operator” error. So many people say that a camera does not focus properly and lots of time, it is the photographer and not the camera. We printed black and white diagonal lines on a canvas that was in turn glued on a board. The board was then mounted on the bulbar of our Landrover. My wife and a friend then drove at 40km/h towards me whilst I photographed the board. The first image was taken from around 70m from the camera and the last image will be around 6m from the camera. The tests were done with three lenses namely a Canon 400 f2.8 L IS USM and the 600 f4.0 L IS USM and the 100-400 f4-f5.6 L IS USM lens at 400mm to see how the camera copes with a maximum aperture of f5.6. The camera was supported on a Gitzo 1648 Carbon Fiber tripod with a Wimberley. The camera settings were as follows:
Before I come to the results I need to explain the different Custom Functions that were used:
See below an example of the test, once again we are not going to show all the results because we took over 1750 images.
After the first test, and that was with the old Firmware 1.04 we were rather disappointed with the cameras tracking ability. The results were better than the MkIII’s results but not a massive improvement. We then loaded the new firmware 1.06 and did all the tests again. Now Canon’s statement about the new firmware and I quote “will improve the camera's performance when tracking receding subjects and subjects that are approaching the camera slowly”. I did not think our subject is “approaching the camera slowly”, in fact we chose the 40km/h approach speed because Canon claims and I quote again “with an EF 300mm f2.8L IS USM lens, the EOS-1D Mark IV can track a subject approaching at 50 km/h up to about 8 meters”, so we thought using the 400 lens we will reduce the speed. The results of the next set of tests were much better, but we still got mixed results. We then analyzed the results and with all the different custom function combinations and then selected the two best settings. Here is then the final results and remember the percentages we have allocated to the tests is only for our specific test and not a general statement. The camera also struggled to focus accurately on the test board when the vehicle was still far from us (70 – 80 meters), but once it locked on the test board from about 50 meters the results were consistently very high. See below the results for the entire distance 80 meters – 8 meters.
As you can see, some major accuracy problems with the 100-400 f4.0-5.6 L IS USM lens with both Custom Setting combinations and with the 600 f4.0 L IS USM lens when we used Left/Right expansion. Why, we have no idea!! The next step is to do the same tests with 1.4x converter on the 400 f2.8 L IS USM lens and on the 600 f4.0 L IS USM.
The above numbers are meaningless if you do not compare them so here are the results of the 1DMkIII and we did only use the 400 f2.8 L IS USM lens, results as with the other lenses results were even worse:
So in conclusion, there is a dramatic improvement in AI Servo focus performance on the MkIV over the MkIII, but I need to mention, when we used the old 1D MkII, yes MkII not the MkIIN, the MkII performed at the same level as the MkIV!!!
I am not going to say much about the 1DMkIV resolution; I will let the images do the talking. At low ISO’s in comparison with the 7D and 5DMKII it is very good, yes the 5DMkII is better, BUT when you move to anything above ISO 1600 the MkIV kicks .... big time!!!
We have not tested a flash to see how the new Auto Flash algorithm works but will do so shortly and add our views to this document. Canon also improved the buttons on the body, see image below.
One problem that we encountered with the new buttons were on the vertical grip, when you are photographing in landscape format ,the outside of your hand can accidently press against the AE lock button on the vertical grip and, because the button now stands further out with a shorter stroke, it easily locks your exposure whilst you’re photographing, especially when tracking a subject from right to left. This happened several times to me in one morning tracking a Fish Eagle fishing.
Obviously we are going to do more field tests, but our first impression of the camera is that it still has some focusing issues on static subjects; the tracking focus has improved dramatically if you compare it against the 1D MkIII, but it is not better than the old 1D MkII. The most impressive about the camera is its noise on ISO’s up to ISO 12800. If you are a 1D MkIII owner I will not hesitate in recommending the 1D MkIV. If you are a 1D MkII or MkIIN owner you will be blown away by the resolution and ISO’s of the MkIV. If you are waiting for the ultimate camera, meaning perfect focus, exceptional resolution and low low noise then you maybe need to hang on for a little longer, because the MkIV is not that yet!!