Monday, November 5, 2012

Nikon D5200 Launch Impresions

And now finally, a DX camera update from Nikon. It's been a long DX drought, cause célèbre for Thom Hogan, and a long period of inactivity for photography bloggers who specialize in the middle ground (yours truly). There's a reason why this blog is "mostly photography"... Nikon made it easy to talk about other things during the DX lull. DX is not dead,of course. It was just waiting to see if you would have bought a D600 instead. Of course, this wasn't the camera launch that the enthusiasts and semi-pros were hoping for, but an important camera for Nikon nonetheless.The D5200 is impressively spec'd, basically taking a D3200-like sensor (but with upgrades in the imaging system), and pairing it with the D7000's 39-point autofocus system and more advanced exposure meter. This is a lot of sophistication for a camera in this segment of the market.In fact, I find it a disappointment that the shape of the camera body is so similar to the D5100. The sum of these parts make this a significantly greater shooting experience than the D5100.

I want to stress "D3200-like" in describing the sensor. Nikon's literature describes it as"new", though they've not always been transparent in describing their sensor origins. (In fact, the never are, you only learn about the sensor after the fact after the whole internet has dug around.) Nikon's literature describes the D3200 sensor as "24.2" megapixels, which we now know as being different from the 24mp sensor used in the Sony NEX-7. They're using "24.1" to describe the D5200 sensor. Depending on how the chip is utilizes, this could still be the same sensor, but the pixel count isn't the tip off that this is something new (or at the very least very upgraded): it's the video specs that do that. The D5200 can now do 1080i at 60fps, all in all, it seems very similar to the NEX-7 chip. Nonetheless, that's still a lot of data to move off of the chip; it's going through the same EXPEED 3 processor that Nikon uses in the full frame camera line. The addition of the WR-R10/WR-T10 radio-based wireless system further pushes this class of camera into more and more serious videography territory.

Back to the still pictures. First we had the mega-pixel wars, when people were chasing more and more pixels. Then we had the anti-megapixel wars, where people were chasing more and more high-ISO noise quality.  The manufacturers are pushing us down the megapixel path again, and predictably, the enthusiasts are resisting. The 24.2mp Nikon sensor in the D3200, and for that matter, the 24mp Sony sensor in the NEX-7 are pretty much state of the art. D7000 and above shooters tend to poo-poo these sensors because they're housed in lower-spec cameras, but if you can get by the bias, the output from these cameras are superb. The objective tests say that the 24mp DX sensors aren't as clean as the venerable 16mp one, but they are close, and the combination of added resolution brings them up to par. If you down-sample a 24mp image from the D3200 to the 16mp size of the D7000, you'd be hard pressed to tell the difference, except that the D3200 would be more detailed. Many enthusiasts disregard or forget about the colour information that gets lost to the Bayer filter; imaging is like anything else, if you sample at a higher frequency, you have more quality information to work with. It is only above ISO 1600 that the D7000 has a clear advantage over the D3200, and I'd be very interested in seeing what the objective tests say about the D5200. So time marches on. For the same technology, generally speaking, yes, less pixels means better per-pixel noise control, but there is a grey area, because more pixels also means more detail. If you can remember that the stated goal of having less noise is to preserve more detail, then you should be happy about having more pixels, and thus more detail. However, you can have your cake and eat it too, if only you wait, as time marches on and technology improves. Nikon has been good at meeting or exceeding expectations with each new sensor generation, so on this front, I am mostly confident that we will see improvements in still picture quality compared to the D3200.

About those pixels, though... I suspect that as the pixel counts increases, many people will be tempted to forgo telephoto-zoom purchases in favor of simply croping into a photo taken with the kit lens. This has already happened with the megapixel war; Nikon is fairly transparent about how many camera bodies and lenses that they sell, but not so much with flash units. I suspect that flash units have been a dying breed among consumers and hobbiests, who are increasingly relying on the high ISO capabilities of the camera body alone. If it happened with high ISO gains, then something similar is going to happen with megapixels. In either case, it's the wrong thing to do if you want quality pictures. With regards to megapixles and cropping, here's why:

These are the magnification increases (decreasing angle of view in degrees) as you move up through the commonly used focal lengths:

Focal Length (DX) in mm Angle of View Percentage Increase

10 109
16 83 31%
24 61 36%
35 44 39%
50 32 38%
85 19 68%
120 14 36%
200 8 75%
300 5 60%

Here's how the linear resolution increases as you move up through the megapixels, using the megapixel counts from all of the Nikon dSLR's. Linear resolution is the measure of of the number of rows and columns; I've picked the width of the sensor, which indicates the number of columns that sensor can produce. Linear resolution is a more useful way of quantifying increases in image quality. To see why, image a two pixel by two pixel image; the total number of pixels is four. If you double the dimensions to make a four by four image, the area increases to 16. So in other words, doubling

Megapixels Linear Resolution (X-axis) Percentage Increase

6 3000 -
8 3464 15.5%
10 3873 11.8%
12 4243 9.5%
16 4899 15.5%
24 6000 22.5%
32 6928 15.5%

In going from the D5100 to the D5200, the increase in linear resolution is approximately 22%. To put that in perspective, the D3200 was the largest increase in linear resolution between sensor generations in Nikon dSLR history. However, compare that against the first table and chart showing how the benefits with increasing focal length scale. Suffice to say, cropping is not an effective solution for replicating a long zoom lens. It is a tool, but but it's not the best one in toolbox. I would also be very careful cropping with a sensor like this. From what I've seen, on  per-picture basis, the noise level is similar to the 16mp sensor. However, on a per-pixel basis, it is noisier. In other words, noise isn't going to matter as much if you have a full set of lenses and are disciplined about your shooting, but it will matter if you tend to crop a lot.

Getting back to the camera itself. The increase in maximum frame rate is welcome (and also implies a boost in memory band-width and image processing power inside the camera). The target audience for this camera will appreciate that, even if it's not always used well. (Anticipation and timing trump fps, and no matter how fast your shutter can click, it won't matter if you you have insufficient shutter speed for your light conditions). I've always likened the D5xxx series as a "dad" camera, since the D40 and it's successors seemed to appeal so well to, well, soccer moms. The D5xxxx, by virtue of the swing out LCD display and the absurdly low price of the D5100 in it's final days has always been a more compelling buy.

That need to be a compelling buy is really what is driving the move to 24mp sensors. With the very capable NEX cameras and the latest m4/3 cameras, dSLR's have to give a reason for why they are higher up the food chain. Ever since the beginning, in the transition from the the six megapixel D70 to the 10 megapixel D80, people have been discounting the need for extra pixels; while there is practical wisdom to this, the simple truth of the fact is that whatever you sell, it's going to have to outsell the other guy. If you plunked the D4 sensor technology into a D5200 sized body, you would have one of the most incredible cameras ever... that would weigh in at 7 megapixels. That's less than an iPhone; that's a hard sell if you have to sell to the fat middle-majority of the market. In fact, I pretty much subscribe to the view than in three to four years time, whatever fills the place of the D5xxx series probably won't be a mirrored camera, but for the time being, the D5200 is an unprecedented amount of camera that is within the reaches of the average consumer. I'm not trying to be all Ken Rockwell here.... it's not the best camera ever, but I'd like you to consider: If you are shooting at base ISO, and took the utmost care, would you be able to tell apart the 24mp images from the D5200 from the D3x? I'd like to remind you that at the time of launch, the D3x was almost 10 times as expensive as the D5200!

What the D5200 does is also raises questions about the impending D400 and presumably D7100. I've always said that the problem with internet chatter is that it's not imaginative enough about what's coming down the pipeline. If you ask anybody on or any other forum like that, they would probably guess at an iteration based on what we know the market to be like now. That sort of thinking would have completely missed the 36mp D800. If the D5200 is like this, the bar will be high for the two cameras above it... And there will be two cameras: Nikon can't fold the D7000 and D300 level cameras together into one model, there's too much revenue potential at stake. These are still two different user groups with two different spending patterns. Sales of the D600 haven't been strong enough to do away with the D300 level camera, and they were never going to do that in any scenario. But based on the D3200 launch, this old rumor about the D400 is looking more and more likely...

No comments:

Post a Comment