EOS 5D + EF 24-105/4L IS USM vs. EOS 40D + EF-S 17-85/4-5.6IS USM

In this test I thought I'd compare two "walking around" outfits. By "walking around" I mean using a single lens that covers the wide to short telephoto range, preferably with Image Stabilization. It's the sort of lens you might carry with you on the camera if you didn't quite know what you wanted to shoot, for example while walking around a city on vacation or taking a walk along the seashore.

With the Canon EOS 5D, the obvious choice is the EF 24-105/4L IS. It covers a good range of focal lengths, it's fairly fast at a constant f4, it's an "L" series lens so its quality should match the EOS 5D and it has image stabilization.

For the EOS 40D, the obvious choice was the EF-S 17-85/3-5.6IS. This gives the same angular field of view coverage as a 27-136mm lens would on a full frame camera like the EOS 5D, so it doesn't give quite such as wide field of view as the 24-105 on the 5D, but it has a longer telephoto reach. Again it's a lens you might carry around when you weren't quite sure what you'd be photographing but wanted to cover a fairly wide range of focal lengths.

All the test images were made with the cameras mounted on the same tripod at the same distance from the target. Both cameras were set to ISO 100 and IS was turned off on both lenses. The lighting was direct sunlight so shutter speeds were easily high enough to avoid any blurring due to camera movement (even handheld, never mind on a Gitzo carbon fiber tripod and B1 ball head!). All the image shown are unsharpened 100% crops from the original JPEG files.

At the Telephoto end of things

Fist let's look at what things look like at the long end of the EF 24-105/4L IS on the EOS 5D. To compare the shot with the EOS 40D, the EF-S 17-85/4-5.6IS needs to be set to 66mm. At this setting it gives the same view as the 5D at 105mm when both cameras shoot from the same distance.

This is what thing look like in the center of the image at f8. Both camera/lens combinations are pretty sharp. The 5D image outresolves the 40D image, but that's pretty much what you'd expect from a full frame/aps-c comparison.

Now let's look at the corner of the image. The 24-105 is designed for full frame and the 17-85 is designed for APS-C, so they're both being taxed quite hard to maintain image quality in the corners (which is at the edge of their respective image circles).

What we can see here is the superiority of the "L" series 24-105 lens, togther with the higher resolving power of the full frame sensor. I was actually a bit surprised to see the extent of the difference, but I looked at images shot at f5.6 and f11 and saw pretty much the same thing (slightly worse at f5.6, slightly better at f11.

In the mid-range

So what about mid-range, say 50mm on the EOS 5D + 24-105 and 31mm on the 40D + 17-85. Below is the shot from the center of the image:

Again both images are quite sharp, with the expected edge going to the 5D based mainly on the full frame sensor and higher pixel count. Would the corner sharpness hold up in the mid range? See below:

The answer is no. As you can see, the 40D + 17-85 image is soft in the corner, the contrast is down and there's evidence of chromatic aberration. Again the quality of the "L" series lens comes out as the difference here is more than can be explained by differences in pixel count and sensor size.

Let's go Wide

So now let's zoom out to as wide as we can while stillmaintaining the same field of view on both cameras. That means the 17-85 has to be zoomed all the way out to 17mm. To match that on the 5D, the 24-105 need to be set to around 27mm. At the wide end of their range both lenses show significant barrel distortion (the 17-85 showing a little more than the 24-105), so I corrected that before making the comparisons.

First let's look at the center of the image again. These are images shot at f8.

As expected, both look good, with the 5D + 24-105 combination looking better. Now for the corners (after distortion correction):

As you can see, even the "L" lens is starting to show a little chromatic aberration, though the overall sharpness is still good. The EF-S 17-85 is struggling hard and shows quite a bit of chromatic aberration as well as a loss of sharpness and contrast. CA can be corrected by using the Canon DPP RAW converter and shooting RAW, or it can be corrected in photoshop. However that still won't help the sharpness of the EF-S lens a great deal.

I did shoot more test images than I'm showing here, at a variety of apertures, but the general conclusions from the images I'm not showing here were pretty much the same as above.

Conclusions

Well, it's clear that the more you pay, the better image quality you get! While the difference in the center of the image isn't all that great, the EOS 5D image is clearly higher resolution. The difference in the corners is significantly greater, with the "L" lens and the 5D being significantly sharper.

You could say "Oh well, just put the 24-105/4L on the 40D then". That's fine, but you no longer have a wideangle zoom. At 24mm on an APS-C camera, you have the same field of view as a 38mm lens on a full frame camera. That's not very wide. In fact be some definitions, it's very close to a "normal" lens (which would be 43mm based on the image diagonal). If you wanted an "L" lens on the 40D, you'd have to go with the EF 17-40/4L - but that doesn't give you much of a telephoto reach. You could pick the EF-S 17-55/2.8IS, but it's not an "L" lens and it still doesn't give you a view either as wide as the 24 or as long as the 105 yo get with the 24-105 on the 5D.

Of course cost is an issue. The EOS 5D + 24-105/4L IS is currently $2600 (until July 19th when the rebates end) which is a saving of around $650 over the normal price of the camera and lens added togther. The EOS 40D is around $900 (until July 19th) and the EF-S 17-85 is around $500, making the pair around $1400, a savings of $200 over the normal price on the camera.

Remember the Canon "instant rebate" program ends on July 19th 2008.

Is the 5D/24-105L combination worth the extra $1200? If you want the highest image quality, then it is, especially if you intend to make large prints from uncropped images. If you're mainly interested in center image quality and make mostly smaller and/or cropped prints, the 40D/17-85 combination may be good enough. I use both combinations myself. Which one I use depends on the circumstances I'm shooting under. For example, the 5D yields higher quality while the 40D has a faster frame rate. For a more detailed 5D vs. 40D comparison see 5D or 40D? Which one to choose.