I recently took a few shots around Lake Monger in Perth, and noticed that some of the landscape shots weren't perhaps quite as sharp as I'd like. In fact, I convinced myself that the old 4 mega pixel S40 was actually better at taking these sorts of shots. Surely some mistake? I set out to see.

I asked on the rec.photo.digital newsgroup and got a variey of answers. Maybe the lens wasn't quite focusing correctly? Maybe the S40 does a lot more image sharpening in the camera? Maybe mirror slap in the DLSR causes a bit more blurriness? Maybe the fact there's more pixels in the 20D shots makes blur more apparent. Maybe my demands were just too high? Hmmm... Some sort of test is in order.

I put the 20D on a tripod which I weighed down by suspending a large rock underneath it. It wasn't a windy day, but I wanted to make the tripod as solid as possible. I set mirror lockup to eliminate mirror slap, then took a test shot of a house some way away at the lens's sweet spot: f5.6 at 1/2000sec. Here's a 200% blowup of the centre of that photo for reference, and below it a 200% blowup of the same shot with the S40. Below that is the S40 shot at 100%:

That looks fine to me. Considering that house is probably a kilometre away, I can't complain about the sharpness of the 20D image. I actually took that 20D shot at ISO 400, but I don't think it made a difference. I'm pretty sure the lens is not misbehaving, and I haven't had any reason to doubt it in the weeks I've had it. It's also obvious that thinking the S40 produces better images in these circumstances is nonsense.

So, back to the 20D. I stopped the lens down so I could get a shot at 1/25sec. I had to go right down to f22, which means the image is going to be blurrier. Here it is, again, from the tripod:

Obviously much softer, as you'd expect. Next I took the camera off the tripod and used the same settings for a handheld shot. I actually reeled off half a dozen. Here's the best of the bunch:

Hmmm, what do we think? The handheld one is slightly worse, but only just. At full size there's not much difference. But, on the other hand, here's the worst of the handheld shots:

which is obviously a lot worse. I honestly did try to keep all the handheld shots the same, relaxing and resetting myself between each one.

Conclusion? There's nothing wrong with my lens, and at slow shutter speeds a tripod makes life a lot easier. It is possible for me to hold the camera steady enough at 1/25sec to get a shot which is very close to the sharpness of a tripod shot, but it's a lottery. The next shot I take handheld, while theortically it should turn out the same as the previous one, might be significantly worse. I put the perceived blurriness of the Lake Monger shots down to unrealistic expectations. When taking photos of buildings a couple of km away, the resolving power of the lens is probably the main limiting factor. And my thinking the S40 does better? That was undoubtedly my imagination.

This seems a really crappy test, but in fact I did quite a lot more research into this. I just couldn't be bothered to write it up! Here's some of the highlights:

• I can actually get reasonably close to tripod stability when handholding with the lens set to 17mm at speeds as low as 1/10th. Most times I get a decent shot. At 40mm and 1/10th I never get a usable shot.
• At 1/15th sec and 40mm I got one halfway decent handheld shot out of 5.
• At 1/50th sec and both 17mm and 40mm I consistently get shots which are close to tripod quality. You can tell they're not as good, but the difference is small.
• At 1/100th it's hard to tell the difference between tripod and handheld shots, even at 200% magnification and 40mm.

So here are the rules for the 17-40mm lens: faster than 1/100th, don't worry about it. 1/50th to 1/100th, pay attention, try to keep as stable as possible. 1/10th to 1/50th keep zoomed out to get a useable hand held shot. If I need to zoom in at slower than 1/50th, I really need a tripod. So the old rule of thumb that says use a tripod for shutter speeds slower than 1/focallength seems to apply.