Canon EOS 20D

Canon EOS 20D
First of all I’d like to apologise for being late in providing this review, unfortunately we were unable to get hold of a camera to test! Since its’ release the 20D has been breaking all sales records so I was keen to get hold of one. Primarily seen as a replacement for the excellent 10D, the 20D has a host of improvements which looked to make it a bargain at the price. Due to time constraints and impending trips I had to pack the 20D test into a couple of days, so I made sure that I used it alongside the new 1Ds MKII which I was testing at the same time. For all of these tests I downloaded and installed the latest firmware for the 20D, from
Technical Specifications and Highlights
Ok, let’s first take a look at the main technical specifications and feature highlights:
New Features / Enhancements
• 8.2 million pixel CMOS sensor with improved low pass filter (less noise) and smaller pixel size (for better colour rendition and lower noise)
• JPEG Large and RAW images are 3504 x 2336 pixels approx.
• 5 fps motordrive
• Buffer size – 6 RAW and 20 JPEG LARGE images before lockout
• Improved startup time over 10D
• DIGIC II Image processor
• Aspect ratio 3:2
• 9 point autofocus with wider viewfinder coverage (quick navigation with new multi-controller button on rear of camera)
• 35 zone TTL full aperture metering with several options
• Shutter speeds 1/8000th sec. to 30 sec.
• ISO 100-1600 with option for 3200
• E-TTL II distance-linked flash capability with EOS Speedlite (untested)
• USB 2.0 socket for high speed connection to workstation
• Monochrome shooting mode with filter effects (untested)
• Support for 2GB and larger CF cards
• Digital Photo Professional (Pony) v1.1 software
• Plus lots of extra bits that don’t matter a jot to most of us
The autofocus on the 10D was a little ropey at times but the 20D is much improved. In fact that doesn’t really do it justice, it is really very good indeed. As usual I tried it on several subjects, both static and moving like the clappers, and for both it coped well. Of course with the static subjects I would expect good performance but using AI SERVO to track moving subjects is usually another matter entirely. But the 20D did the job, allowing me to nail some reasonable flight shots will little effort. As usual it missed a few, which I would expect, but once locked on it stayed locked on and that is all we can ask of it. With this swan image I tracked the swans for 30 seconds before they landed and felt confident that the 20D had the lock all the time.

I found the expanded autofocus point range very useful and liked the fact that I can now reach farther towards the edges of the frame. The autofocus points at the edge also seemed to be very responsive, although I did find that the central ones were the most accurate (as expected).

The one feature of the autofocus that I really HATED was the new multi-controller button. This tiny button sits above the rear thumbwheel and one of its functions is allegedly to move the focussing points around the viewfinder. Absolute pants. Not only is it annoyingly inaccurate at placing the point where you want it to be but it’s small and fiddly. Right idea Canon, wrong implementation.

Apart from that one gripe however I found that the autofocus was extremely good and would trust it in any situation that I might find myself in

Buffer Considerations

Before seeing the specifications for the 20D I had hoped for a much larger increase in buffer size. The problem for Canon of course is that the 20D now has a much larger file (8MB TIF compared to 6MB from the 10D) and it takes time. In the confines of my studio I set about checking out the timed performance of the buffer; the results are subjective of course and will not be 100% accurate but at least they will give you an idea. For the purposes of the test I used an Integral I-PRO 2GB card and the camera was set to RAW:

Time after lockout when 1 shot available – 0.8 / 1 second
Time after lockout when 5 shots available – 3 / 3.3 seconds
Time after lockout when buffer cleared – 5 / 5.5 seconds

Clearly Canon have worked on the buffer speed and I was particularly impressed by the reduced delay after lockout. One thing that became apparent during my testing, both field and studio, was that after lockout the 20D continuously freed up a single shot even though the shutter was permanently pressed. In other words I could not get the 20D to the total lockout stage, it always gave me the chance to shoot again (albeit after 1 or so seconds). Of course the JPEG performance is much better as the buffer holds at least 20 JPEG LARGE images before lockout; note the timings above will be much shorter to JPEGs too.

I am still disappointed that the 20D does not have a larger RAW buffer size, and 6 may be too small for those people who specialise in action photography. Still the buffer performance is worthy of praise and for most photographers the performance will be a marked improvement over the 10D.

Shutter Noise

I was surprised about how loud the shutter is on the 20D, it is noticeably louder than any of my other DSLRs. I can find no reason for this but photographers who work with shy subjects or in environments where noise is an issue, should be aware of this.

Oh great, they’ve released a new version of the Digital Photo Pony, marvellous. I won’t waste valuable column inches on this (read my 1D MKII review for further info) but will suffice to say that it is just as crap as the last one. No workflow and questionable output quality. I spoke to a couple of the independents and thank god they’ll have support out soon. Come on Canon, surely producing good software isn’t that hard?
Image Quality
As a professional I can work with low buffer sizes and dodgy autofocus controls but the one factor that I cannot tolerate is poor quality. To assess the quality of the 20D files I used the camera in a variety of light situations and processed the resulting RAW images with no sharpening or noise reduction. Then I magnified each upto 100% and went over the section of the image which I knew would highlight any quality issues. The one limiting factor was the Canon software, and I am sure that with any other software the results would be better still.
Detail Test

With the new image processing features the 20D promises a significant quality improvement over the 10D. For this test I shot several different subjects at ISO 100, worrying more about getting fur / feathers in the shot than any award winning compositions. The squirrel image here is perhaps the easiest image to check out the detail.


Squirrel Close Up

Cropping in tight to the nose and whiskers, the detail is excellent considering there is 0 sharpening applied to the image. In many ways it looks very similar to the images that I have been getting from my 1D MKII, with plenty of detail straight from the camera. Of course it doesn’t leap out at you like the 10D, since the new anti-alias filter is thicker (to reduce noise) and will make the image appear slightly soft. But a tiny amount of sharpening brings out all the detail and is plenty enough for even the most picky photographers.

It is interesting to note also that at higher ISOs (see below) the detail level is even more astounding.

High ISO Test
On a very gloomy, dark day I decided to test the 20D’s ability to work at high ISO values. Now rarely would I even consider shooting about ISO 200 so for this test I stretched my limits and set the 20D to ISO 400. Taking into account the conditions, this would be a very good test indeed. My subjects were some local wild goats, and again I went for a good test image rather than stunning composition so apologies in advance.
Goat Close Up
Cropping in tight on the goat’s ear, there is very little noise evident in the diffuse background and a fine amount of detail in the fur. To me this looks just the same as an ISO 100 image and I would have no hesitation in sending it to any of my clients. Even brightening the image and bringing out a little shadow detail, the noise is still conspicuous by its absence. I actually shot the swan image (see above) on ISO 400 and did notice some noise apparent in the background, although these kind of skies always show the noise more than others.
Shadow Test


Noise manifests itself most in dark shadowy areas of an image, so hopefully this sunset shot should show any up.

Sunset Colse Up

I’ve cropped very tightly into the image, to show the sky, water and land. None of these areas has any noise; I’d expect there to be some at the transition of the light / dark areas but there is none.
Quality – Conclusion
Although my test was quite limited I did manage to test the 20D in a range of situations at varying light levels. Each time I was pleased with the quality and only on a couple of occasions did I see any evidence of noise. Of course I didn’t shoot at really high ISO levels of 1600, where I am sure there would be some noise, but even so I suspect the performance would still be superb.
Colour Balance
The 10D definitely had a tendency to give images with a slight orange / red bias. I found the 20D, when left to it’s own devices on AUTO WB, gave good results most of the time. WB Cloudy was very orange, with WB Shade less so and acceptable for certain images. If you shoot JPEG then this of course is a consideration, for RAW shooters I would suggest always setting to AWB and adjust at the software stage.
Well as expected the 20D scored very highly with my tests. In fact the only negatives I found were the shutter noise, autofocus controls and the software. Admittedly I would like to see a bigger buffer too, but that is what we are provided with and that’s that.

For new DSLR users, or those upgrading from D30 / D60’s I would say that the 20D will show a huge benefit. Of course it is being heralded as the replacement for the 10D and the obvious question arises whether it is worth you upgrading from the 10D. Well IMHO the 20D does give better noise performance, image quality and buffer performance than the 10D, plus it starts up instantaneously. Many photographers that I have spoken too over the Christmas period have expressed concern as to whether a 2MP increase in sensor size over the 10D will make any real difference to them. It’s certainly been my experience that with the larger number of pixels, smaller pixel size, new DIGIC II processor and anti-alias filter, the 20D produces a much better image than the 2MP difference would at first suggest. So on balance I would say that if you’ve got the money and want to upgrade then the 20D would be a definite improvement over the 10D.