I’m not much of a pixel-peeper, and I tend to shrug and yawn at in-depth technical reviews of the latest DSLRs and their performance. Nearly any current prosumer or “professional” DSLR is capable of taking amazing pictures.
While I admit I’m a bit of a gear head (I recently upgraded from the original 5D to the 5D2), I find esoteric discussions on the hyperfocal distance, circle of confusion, etc. rather boring. Understanding these topics is important at some level, but I don’t think they contribute mightily to my ability to make nice photographs.
Noise performance, and how it changes with ISO setting is one of those topics that is important, though. A recent post by my friend and fellow photographer Scott Lawson got me thinking about this, and I decided to set aside some time to run these tests on my 5D2. I’m sure these tests have already been performed, but there is something educational about walking through the steps yourself. Every camera is different, so you might find it useful to try this out on your own equipment.
I placed my tripod at the top of the stairs, set the camera to aperture-priority mode (f/5.6), and then used a cable release to take a succession of identical pictures (RAW files). I started at ISO 50 and continued through ISO 100-160-200-320-400-640-800. I stopped at ISO 800 since that’s the upper limit of what I normally use in the field, though the camera goes much, much higher.
If you’re familiar with ISO, you know that each jump (i.e. 100 to 200, 200 to 400, and so on) is equivalent to a full stop. Canon cameras apparently offer intermediate ISO values such as 160, 320, and 640, which I suppose represent fractional increases in these stop values. I found it interesting that the shutter speed didn’t change from ISO 160 to 200, or from 320 to 400. There must be some sort of in-camera exposure compensation going on, and these fractional values are probably just a marketing trick to advertise increased ISO range. It’s hard to tell, but I think I can see a tiny increase in noise between ISO 640 to ISO 800, even though the aperture and shutter speed is identical. Is this due to in-camera noise reduction? I guess that’s a question for Chuck Westfall (tech guru at Canon USA).
First, I imported all eight pictures into Photoshop and adjusted the white balance, which was way off. I have a 400 gallon reef aquarium, and it casts a bright blue light throughout the house. The actinic lighting on the aquarium tends to wreak havoc on the 5D2’s auto settings.
Next, I selected an area in shadow (where noise tends to be worst) and made identical 100% crops of each image. This allowed me to easily compare the noise at different settings.
In the field, I usually shoot at ISO 100, 200, or 400, depending on conditions. Noise increases with ISO, and it’s generally recommended that you use as low an ISO setting as possible. On a newer camera like the 5D2, I’ve been working under the assumption that noise performance is still very good at ISO 400, so if there is ANY worry of wind blur or camera shake, I’ll just dial it up. Yes, this results in more noise, but those extra two stops can really help to keep things sharp. If I’m shooting a tree on a windy day, or my tripod is standing in a flowing river, I think it’s a fair tradeoff.
In fact, when I’m shooting wildlife or other moving objects (like surfers) I sometimes dial ISO as high as 1600 if conditions warrant. Things start to get a little noisy at that level, but sometimes that’s the only way you can freeze motion. Obtaining sharp focus is almost always more important than avoiding noise.
My point is this; there is no hard and fast rule when it comes to selecting an appropriate ISO setting. Many books and online tutorials will tell you to use the lowest possible ISO, but there are plenty of circumstances where you should break that rule. Don’t be afraid to take advantage of your camera’s technology; it exists for a reason.
So, let’s examine the results. These are unedited 100% crops from my 5D2:
To my eye, ISO 50 and 100 are virtually identical (just a slight edge goes to ISO 50). Noise starts to increase at 200 and is quite noticeable by 800. Remember, these are 100% crops and we’re pixel-peeping here. If this was a 16″ x 24″ print, these crops would represent an area about 1.5″ x 1.0″ in size. That’s pretty tiny.
More important, check out the difference in highlight detail between ISO 50 and 100. I’ve read that ISO 50 will clip the highlight detail by a full stop, and this appears to be true. There doesn’t appear to be any benefit to shooting ISO 50, and losing that detail at the bright end is actually pretty concerning. Again, I wonder why Canon bothers offering this ISO expansion, if it actually decreases image quality. Go figure.
You might be able to see the clipped highlights in the first picture. Identifying this sort of thing becomes easier with practice. If you examine the brightness histogram for each image, you can see that the image shot at ISO 50 is clipped at the right.
I’ve often read that noise tends to hang out in the blue channel, and I confirmed this by examining and comparing the individual color channels. I won’t post the images here, but suffice to say that the blue channel was the most noisy by far. If you use noise reduction software as part of your digital workflow, this is important to keep in mind.
At the end of the day, I feel confident that noise performance on the 5D2 is excellent. If making small prints or viewing images on a computer monitor, I don’t think noise is an issue whatsoever, even as high as ISO 1600. At larger print sizes, I believe ISO 400 is a perfectly viable option, and ISO 800 is acceptable if you include some sort of noise reduction in your workflow.
This is important to keep in mind, because many of us who shoot landscape photography tend to get wrapped up in the minutiae required to obtain sharp images. We use sturdy tripods, mirror-lockup, cable releases… the list goes on and on. We take these steps so that it’s possible to make large prints, if we so choose. It isn’t necessary, and sometimes it can be detrimental, if you follow these same rules and habits in all photographic situations, though.
When I’m shooting a casual family get together, or even a special occasion where I know my photos won’t be printed at 20″ x 30″, I will gladly use ISO 1600 (or higher) to help get the shot. If I’m shooting indoors and without a flash, what sense does it make to sacrifice a sharp image for the sake of noise – noise that you won’t ever see because the pictures are destined for a web album or as 4″ x 6″ prints?