Comparison: Nikon D700 and Olympus E-3 in Low Light

Nikon: 4 out of 5
Olympus: 3 out of 5
Yeah, but: What are numbers when we have pictures?

The Long Version: Go on any internet camera forum, and there will be 'what should I buy?' questions. Someone will start a thread with a message that reads "I'm choosing between X and Y, and would like insightful comments from people with experience using these two thingies." It's guaranteed that someone, usually the first person to reply, will start their message with "I've never used either thingy, but _________".

My mother always said, "Why would I want the opinion of someone dumber than me?"

So with that in mind, here's my disclaimer:

I've never claimed to be an expert on anything in this blog: if you do a google search on "reviews of dubious consistency", thewsreviews is ranked second. (Disappointing, actually, I was hoping for first place.) I'm just a guy who uses various thingies and writes about them. Secondly, I don't generally "test" anything. I use things. In this case, I've compared these two cameras with the lenses that I have in the way that I usually use them. Different people will differ, and that's not anyone's fault. (My other blog has almost five years worth of photos - 600+ posts - and is almost entirely composed of hand-held static subject photography.) People who need to freeze action, or who are smart enough to use a tripod, don't need to read any further.

Finally, this whole experiment is personal and the interpretation of the results is subjective, so don't expect any Authoritative Numerical Values (ANV™) to neatly summarize the conclusion.

Meet the Contestants:

The Nikon D700 is currently DxOMark's top-rated camera for low-light ISO performance, with an ANV™ of 2303. I doubt that many people would dispute that it's the camera with the lowest noise on the market right now.

The Olympus E-3 is about a year older than the D700, and has an ANV™ of 571 for low-light ISO from DxOMark, putting it in 37th place - tied with the Canon 10D, but slightly ahead of the Nikon D60. Some people might argue with that, but an equal number would probably be surprised that it ranks that highly. Four-thirds ("4/3") format cameras have traditionally had higher noise than the slightly larger 'cropped' cameras. What the E-3 offers is in-body image stabilization, which I will be abbreviating as IBIS to make this review go faster.

My goal is to get the best results for low-light hand-held photography for static subjects from two different technologies, high-ISO and IBIS. My best autofocus Nikkor is the 85mm f/1.8, so I matched it with the Olympus 50mm f/2.0 macro lens. They're nearly the same weight, have a similar field of view, and B&H even sells them for the same price. The E-3 came out to 1.22 kilos, while the D700 checked in at 1.47kg. It's interesting to note that while the lenses cost the same amount, the difference in the body price means that I could buy two of the E-3+50mm kits for less than what the D700+85mm costs at B&H right now, but it's not really relevant to how I evaluated their results.

The Setup:

I shot two different angles on an abandoned car in an underground parking garage. The lighting is consistent throughout the test, and each camera had a custom white balance set. Both were set to continuous shooting at five frames per second, and I kept the second photo from each setting. Both were only using their single central AF point. I stepped through the apertures from wide open until I was getting shutter speeds around 1/5 of a second, bumped the ISO up one stop, and then repeated until I maxed out the ISO on both cameras. For the Nikon, I shot at f/1.8 and f/2.2 before picking up the traditional whole stops at f/2.8; for the Olympus I just shot the whole apertures.

The tail-light test was shot from about 20-25 feet away from the subject, using my best hand-holding technique while standing upright and unsupported. I set my Sekonic meter to a shutter speed of 1/100, which is about the reciprocal of the focal length, and ISO1600. It told me that I should shoot at f/1.4. My focusing point is the yellow turn signal lens.

For the second test, I was leaning against a pillar that put me 10-12 feet from the subject. My Sekonic-metered exposure was 1/100, iso1600, and f/3.6 - eight-thirds of a stop brighter than the tail-light test. The focusing point is the lock in the door handle.

The Cull:

I took many, many photos. Once I'd reduced them to only the second frames, I made another three passes to select the best images. For the first pass, I magnified the images to 100%, centred on the focus point, and selected them based on sharpness. I was surprised to see that I cut all of the shots taken with the Olympus 50mm f/2.0 wide-open: in my years of using this lens, I've never felt that sharpness was lacking. The cutoff for shutter speed with the IBIS-equipped E-3 fell at about one tenth of a second. The D700 wasn't as clear-cut, with some of the photos at f/2.2 being acceptable and some were rejected; likewise some photos taken at 1/30s were acceptable for motion blur, although most needed to be at 1/60 or higher. Naturally, I wasn't looking at the exif data while sorting through them.

For the second pass I was looking only at noise. I set the magnification to 50%, as this is still far larger than web display and more demanding than printed output. I found that I cut all of the E-3 photos taken at iso3200, and all of the D700 photos taken above iso6400. I may have been a little more forgiving with the Olympus, but I also think that Lightroom does a better job converting the D700's noise to monochrome. So count this as a two-stop advantage to the D700, since DxOMark consistently measures the D700's actual sensitivity below its nominal rating. ISO6400 is actually 4871, while the E-3's iso1600 is measured at 1587 - to the D700's 1277.

Finally, once I had photos that I considered good for both sharpness and noise, I went through them again. Rather than judging them only on their own merits, I took only the top six from each test, regardless of which camera they were from. I cropped a section from these in an agnostic 4x5 ratio and printed them out, with the average DPI being in the low-to-mid 300's. Then I took them to the camera store where I work part-time, and let a half-dozen other photographers pick their favourites for detail, sharpness, and noise.

The Results: Tail-Light Test

Nikon D700, 1/25, f/4.0, iso1600

The photo above was consistently picked as the best for sharpness, detail, and noise. It's not a surprise that it was shot with the Nikon. I didn't tell my judges which was from what camera, and even I didn't know the shooting settings for each one, but these are camera-savvy people and could usually figure it out.

Olympus E-3, 1/13, f/2.8, iso400

Coming in second overall, but with less enthusiasm from the judges, is this shot from the Olympus E-3. The differences are obvious: darker exposure, the tonal range is more compressed, and there's less variation in the colour. Compared to the original scene the Olympus is more accurate, but the Nikon's file is going to allow more modification with less deterioration. From the exposure data we can see that the Oly has a shutter speed one stop slower and an ISO two stops lower, showing that IBIS isn't making as much of a difference as it's supposed to. Theoretically I could get a faster shutter speed or lower sensitivity from the Olympus by opening the lens up one stop for equal depth of field, but I had already rejected the f/2.0 shot for lacking sharpness. This goes back to the witticism that in theory, theory matches reality, but in reality, it doesn't.

My co-worker image evaluators didn't get to see these, but the previous two photos are the full images after being put through Lightroom's Auto Tone feature. It has brought down the D700's highlights, and shows just how little real difference the E-3's extra depth of field makes in the full-sized picture. It may also explain a bit of the metering difference, and shows a how the colours are rendered even after setting the white balance with my Ezybalance. Once again, I'm going to say that colour is more accurate and consistent from the Olympus, but anyone who really cares about colour should be profiling their cameras and not relying on the default Adobe profiles the way these images are. They would also be editing the raw file to get the most out of if, which I haven't done. Look for that in a future comparison.

Two more Nikon photos were mid-pack for the tail-light series. One was 1/50, f/5.6, iso6400; the other was 1/160, f/2.2, iso3200. The other two Olympus photos were considered the least technically adequate, and were shot at 1/25, f/4, iso1600; 1/50, f/2.8, iso1600. These two were pretty much interchangeable as the weakest, so the one above the one shot at f/4. I'll say that the best of the Olympus results are close to the best Nikon image, and there's more variation in the quality within each camera than between the brands. But that on average and in particular, the Nikon does produce superior files.

The Results: Door-Lock Test

This test was less extreme, being both brighter and shot with a bit more support. I expected this to be the place where the Nikon stomped the Olympus, but in a reversal from the first test, it just didn't happen. Overall these photos didn't get as much attention, and two photos were picked as the better ones but without a clear preference between them.

Olympus E-3, 1/40, f/2.8, iso1600

Nikon D700, 1/30, f/2.8, iso1600

Once again the Olympus has metered the scene darker, with a shutter speed 1/3 of a stop faster. Shot at the same aperture, the greater depth of field is obvious in the Oly image; shot at the same iso sensitivity, there's not much difference in noise. The second tier of photos were another Nikon-Olympus tie, and shot at 1/160, f/2.8, iso6400, and 1/15, f/4.0, iso1600. It's not a big leap to guess which camera produced which file. The photos with the lowest approval rating were shot with the D700 at iso12800, one at f/2.2 and 1/400, the other at f/2.8 and 1/160. It seems like the E-3's high-iso results suffer more dramatically in the lowest light, and there's also a clear degradation in the colour and clarity of the D700 above iso6400. In fact, one of my judges singled out the Nikon's photos as inferior because of their lack of colour, even though he mistakenly thought that meant that they were taken with the E-3.

Choosing a Winner:

I'm lucky - I don't need to choose a winner. I own both cameras, and have no vested interest in proving one superior or defending the underdog. I didn't buy the D700 for its high-iso ability, and don't need to freeze motion: I'm a stationary-object kind of guy. Since this is a subjective evaluation I'll leave choosing a winner to each individual, and can even supply the gallery link and password to the full-sized photos if anyone wants to write their own conclusions in a comment.

But I won't cop-out completely - the D700 is better in low-light. Its files also have much greater latitude and can be modified more extensively. There's no surprise there. The surprise is that the E-3 remained competitive, and captured better colour. If I had to take photos that would be well-served by a short telephoto, then I'd immediately reach for the D700. But if I had some low-light needs that would benefit from my Olympus 7-14 or 35-100, I would use my E-3 without hesitation. I'd just keep it at iso1600 or lower, and be more careful with the exposure.

Choosing a different lens, for either camera, would naturally change the results. The previously mentioned 35-100 would make the price of the Olympus kit match the Nikon, and its greater weight makes it easier for me to hold steady. If I had chosen the Nikon 105/2.8 VR Micro lens as a better spiritual match for the Olympus 50/2.0 Macro, then the Nikkor's image stabilization would have completely changed the outcome. While this has been a comparison of ISO vs. IBIS, the combination of the two technologies would be unbeatable. But if a camera that's twice as expensive also needs a lens that's twice as expensive in order to be definitively and unquestionably better than a camera with a sensor that's one-quarter the area and a year older, is that still a win?

If you know that a camera will give you a specific ability that you need, buy it. If you want a particular camera and can afford it, buy it. But if you feel like you need the latest and greatest because it scored better in a test, don't forget to look at the rest of the picture. There may be a whole lot of other things that are as important, and possibly more important, than any one single factor.

And don't forget that the point of having a camera is to take pictures.