Update: March 2010 This article has gotten a lot of use since it was first published on Jun 16, 2009 and I thought that it would be a good time to re-issue this. Especially in light of @fcwestfall's posting of this excellent article “This lens is soft” and other FACTS from LensRentals.com
“My images from the Canon EOS 5D Mark II just aren't as sharp as my xxxx (usually a 1.6 crop camera)”
I can't tell you how many times I've read something like that lately. Well, I think we all know better, the Canon EOS 5D Mark II would be an absolute failure if that were true. But it isn't – people around the world are still clamoring to get their hands on a 5D mk II.
So, why do people say their Canon EOS 5D Mark II stills aren't focused?
I believe it comes down to a wide variety of factors (none of which is a camera failure) and we'll discuss each one in more detail:
- 21mp images – this will eat inferior lenses
- lots of pixels in 21mp (usually compared to 8 or 10mp) – 5616 x 3744 pixels
- High ISO Noise Reduction on by default
- Microadjustment needed on lenses (also complaints that the AutoFocus(AF) misses)
- jpeg vs RAW
21mp images – watch out when using your old lenses
Many people say that the resolution power of the camera exceeds most lenses ability to resolve details – and if you aren't using good glass, you may well think that the camera is lousy – but in fact the issue is the lenses. The lenses you use on the 5D mk ii had better be very high quality or you'll be disappointed when looking at them at 100%. Does that mean you've got to buy all new lenses? Maybe… it depends on how you use the camera. The 5D mk ii shines best with really good glass. Using inferior lenses will disappoint you with this body.
Lots of pixels in 21mp (usually compared to 8 or 10mp) –
I also believe that many of the folks with ‘blurry' images have to do with looking at 21 mp images at 100% on a monitor and they're comparing that to the 12mp of the 5d (or 10mp or less on a 1.6 crop camera) – these images are huge and looking at them at 100% may lead some to think that they aren't in focus. The images from the 5D mk ii are huge. The full dimensions of the 5D2 are: 5616 x 3744 pixels and my 24″ monitor is 1920 x 1200, so the 100% image is almost 3 times larger than my 24 inch monitor. People used to seeing 40D images at 100% will see a lot more details that they never saw before on the 5D mk ii. Sometimes those details at this size would look fuzzy. I'm not saying the 5D mk ii images can't be tack sharp at 100%, they can. But you've got to realize there's a lot lot lot more data in the 5D mk ii images at 100%. Let's not forget that most people (ie. your customer or someone viewing on a web page) will never see these images at 100%! If you print them at 100% they'd be monsters… you'll likely print them at much smaller sizes and when you take 21mp down to a reasonable size, they'll look darn sharp. You can get a little ‘sloppy' at 100% when you're reducing it for print.
FYI, here's a sample image I've taken with the 5D mk ii… on the left is an image at about 8×10 resolution on my monitor – the right side is the 100% blow up… is it “tack sharp” at 100%? nope… (it was handheld on a slightly breezy day) but it looks very nice at print resolutions. It is stunning at 8×10 size.
Here's the image on smugmug, you can see it at the largest size (which is 50% of the original size!)
Looks darn good doesn't it? Does it have to be ‘tack sharp' at 100%? I don't think so – not unless you're printing it on a billboard.
High ISO Noise Reduction on by default
High ISO speed noise reduction – which is in the custom functions (C.Fn II: Image section). The 5D2 manual says (page 176 C.Fn II-2), “Although noise reduction is applied at all ISO speeds, it is particularly effective at high ISO speeds.” There are 3 settings and it is on “Standard” by default. I turned mine to “OFF” as I decided I didn't want the camera applying any NR. If I need it, i'll apply it myself with Noise Ninja or Neat Image. I suggest you try the different settings on the same shot to see what works best for you.
I decided that maybe a quick test might help convince some people that this needs to be set for your personal preferences. Using my 5D mk ii on a tripod, I took a couple of photos of some refrigerator magnets. The results are below (yes, I used ISO 1600 and used the timer at 10 seconds to hopefully remove any shake – all settings were manual and I turned autofocus off after focusing with live view at the 10x setting). The screenshot is of the RAW files in Aperture. It is not jpegs (tho the image itself is a jpeg). The difference here is subtle, but in some of my tests, the results were very significantly different.
Microadjustment needed on lenses (also complaints that the AutoFocus (AF) misses)
People are very happy about the microadjustment feature in the Canon EOS 5D Mark II to help focus specific lenses. Returning to the resolution issue, it is quite possible that your lens on an older camera may have been just slightly missing focus when you were using autofocus, but since the resolutions were smaller, it may not have been obvious. Now, with the huge resolution of the 5D2, you may now notice more images where the AF missed just a bit. You can correct that in the camera with the microadjustment feature!
jpeg vs RAW
Recently, I've been reading more about the compression applied to the jpegs in camera. I almost always shoot in RAW and I use Aperture to manage my images (you can also use Adobe's Lightroom – see the buyer's guide for more). I'm able to do more with the RAW files in terms of post processing and then export to jpeg or png. Jpeg is a compression algorithm, and therefore you'll most likely lose a little bit when the camera stores the jpeg. And, don't forget that most camera makers apply a bit of sharpening to jpg images – so a side by side comparison of a jpg and RAW file might show the jpg as sharper.
So? What's the net?
I've seen enough stunningly great photos from the 5D mk ii to know it isn't the body that is the problem… in all likelihood, it is the settings (High ISO NR) and the lens (that was ok on a smaller body but not at 21mp), or the lens needs microadjusting on the 5D2 or a combination of these things.