Update: I realized today that I approach this sort of article from a perspective that is not shared by everyone, so I should be very clear about why I like to compare cameras that are theoretically interchangeable. Meaning that one can buy an APS-C system for general purpose shooting or one can buy an m4/3 system for the same purpose. Each has advantages and disadvantages like body and/or lens size, size of system (number of lenses for example), video capabilities, image noise, speed of AF, AF tracking, buffer size, EVF versus OVF and so on.
But a lot of people tend to focus on one thing because it is the subject of so many comparisons --- crops at 100% to gauge detail and noise. I prefer to approach this subject to show the differences for the average image when examined in the comparators without a bias towards the nth level of subtlety. In other words, I like to note when images are pretty similar, more of a cup half full philosophy. I firmly believe that technology has advanced to the point where m4/3 and above all make “dLSR quality images” in most lighting conditions.
Yet, when examining crops from these cameras, many people would declare the APS-C sensor the clear victor and wonder why m4/3 even exists in this market, as DigiLloyd did in a recent article. Well, I cannot agree with much of what he said because m4/3 does not suffer from noise at embarrassingly low ISOs unless you are a raging measurebater. So in the end, you may agree with the following or not, but if you do then you are free to choose a system on a far wider set of parameters.
A truly interesting question. These cameras are all in the same ballpark, cost-wise, except for the GX1. The GX1 is very inexpensive if you look around. I got mine for 300 bucks and at that price this thing is the bargain of the century.
So the D7100 review was recently posted at DPReview.com and I urge you to go read it if the state of the art interests you. The D7000 remains a superb camera, and there is certainly no need to rush out and upgrade, but the new camera is pretty magnificent.
What interests me, though, is the distance form APS-C sensors to m4/3 sensor of the modern variety – which I define any anything at 16mp after the GH2. The GH2 was the first generation of 16mp sensors and has a lot of difficulty in shadows, so I define it these other sensors are second generation (GX1) and third generation (E-M5 & GH3.)
I also think these sensors all look very much alike at base ISO and comparing them there is basically a waste of time for RAW shooters. JPEG shooters, on the other hand, are going to want to compare output so that they can choose the color rendition they like. Don’t get too twisted over tone curves and sharpening, as that is always controllable through the user interface. Color rendition too, but only to some extent.
Where I get interested is at extreme ISO, because it tells you where your camera’s limitations set in. If you are at a party, can you set the camera to 6400 ISO and fire away? Or will you ened up with nothing but garbage.
Well, the good news is that you might get something useable. Basically, you are looking for a reasonable level of saturation and definition of lines, and you are looking for a reasonable attempt at low contrast fine detail like hair. So let’s look at these for a moment, using the excellent comparator at DPReview.com …
Here is what the 6400iso JPEG looks like of a copr everyone has seen by now … the little robot in the test scene:
Pretty hard to argue with Olympus’s rather strong processing when applied to high ISO. It wins easily with the D7100 a little behind. The GH3 is a bit further behind and the GX1 brings up the rear. The Olympus JPEG engine is masterful at bringing out details at higher ISO, and JPEG shooters need look no further. Of course, in such lighting, the D7100 will probably spank the OM-D with better AF performance etc, so there are many other considerations in your buying choice. But for now, consider the OM-D as a very strong contender for JPEGs.
RAW shooters have an entirely different decision to make. Here is what that looks like …
Ok, now that;s much more interesting. First, they all look alike, with the D7100 having a nicer noise pattern as would be expected. This makes it obvious that someone with even a modicum of skill will get a better image form the D7100 than from the three m4/3 sensors. But more importantly, there is almost no difference between the 3 m4/3 outputs.
Do remember that, even if we see subtle differences, we cannot really take them as significant, because these are Adobe defaults for the camera in question and anyone who judges by default settings is not thinking very clearly. These are RAW images and they are meant to be processed by the photographer. Please internalize that thought now, else everything I say will seem like it is in Martian.
So we’ve seen that these are all pretty competent in terms of saturation and fine details like those hairs that are on spread across the robot. Not enough difference to even comment on.
But what about a field of these hairs, or maybe a feather? Let’s have a look at a feather …
The Nikon looks pretty decent in RAW. The OM-D and GH3 look a lot alike, and the GX1 is a bit behind those two with a tad more noise. An older sensor, so this makes a bit of sense. But how that small amount of noise really does not hamper the person processing the files as much as you might think. Once you dial up color noise reduction (and in Lightroom 4 / ACR 7 this is just excellent) you really aren’t left with any blotching of consequence. I actually prefer the blotching pattern of the GH3 here over the E-M5, but remember that these are default settings and there could be more or less of each NR technique automatically applied. The only way to know what these really mean is to process them yourself and see what you get.
That last comment is extremely controversial as the vast majority of enthusiasts on the forums believe exactly the opposite. They think that Adobe defaults are the only fair comparison. They are wrong, of course. Practically speaking, no image is ever left at the defaults and so that state is meaningless except for cursory examination.
Now, how about some hair?
Whoah …. the GX1 looks as good as the rest. What is going on? Again, default settings differing slightly. What it does tell us is that the GX1 sensor is very good, even at such rarified ISOs.
There is a lot of whinging on the forums right now over Panasonic’s strategy to reuse sensors. Well, these are great sensors, obviously. The updated GH2 sensor in the G5 and now the G6 has fixed the shadow issues and looks as good as the GX1 sensor. So I have no qualms about the fact that the GH3 sensor remains in the flagship camera, and I have no qualms about the fact that the OM-D sensor is not in the Panasonic cameras. It just does not affect me as a RAW shooter enough to get mu knickers twisted over it.
I mentioned shadows … let’s take a look …
The GX1 is a bit behind here for sure. A bit more processing required, but you will still be able to tame it because it is in the form of more chroma noise. The OM-D and GH3 will have a bit more detail I suspect. When I get some time, I will process them myself and post a second take on this article. But for now, you will want to bone up on your high ISO skills :-)
Bottom line: JPEG shooters should probably buy the D7100 or the OM-D if you want to have a great deal of latitude to 6400 ISO. But remember that professionals manipulate or add light, so they are not troubled by such limitations. Enthusiasts are of either ilk, but if you are really serious about shooting in the dark, you really ought to consider learning about light or using full frame and be done with it.
RAW shooters have more latitude and can probably shoot any of these unless 6400 ISO is a way of life. Then I would recommend full frame again, since it is much cleaner than any of these cameras.
But for most people, such high ISO is a rarity and all of these cameras are going to be pretty excellent. Choose based on ergonomics, features and performance.