Thursday, January 10, 2013

Nikon D5200 High ISO Review Part 2: DX or FX? Comparison with D600

More information is coming in about the D5200. For a camera that was announced weeks ago, it's surprising that so little has been known objectively about its performance...until now. ePhotozine has samples up for the Nikon D5200. For a comparison of how the D5200 compares to other DX cameras, refer to this earlier post. So far I've been happy with all of the sample images that I've seen from this camera. It's not so much that the camera is an improvement in image quality over the previous DX cameras... I would describe it being as the D5200 being able to make 24mp work better on the DX format than the D3200.


First things first, ePhotozine's sample images are more standardized than Photography Blog's so it was interesting to see if the same trends that we saw across the different cameras held true with different testers. For the most part, I would still say that is true. As the ISO increases, the D5200 is cleaner than the D3200, and has a finer texture in the image noise, and has slightly more colour saturation. Compared to either, the D7000 retains edge detail better at high ISO. However, up to ISO 3200 (which is the highest I would prefer to go with either camera in normal shooting conditions), I'm seeing the same trend that I see in Photography Blog's images: the D5200 has similar levels of noise as the D7000/D5100, but with more pixels. Again, based on these samples, I'm comfortable in saying that the D5200 is now the state of the art for Nikon's DX sensors.

Compared to the D600


First things first, ePhotozine's ColorChecker samples seem slightly brighter for the D600 than the D3200. I'm not sure if this is an inconsistency in lighting or if the two cameras meter slightly differently. With the difference in sensor area size (2.3x), you would expect about one stop's difference in terms of image noise, and that's sort of what I'm seeing. It's about that much, in some samples the difference seems slightly less, and in others it seems slightly more. The D600 seems to preserve edge detail and low contrast detail better, but for most subjects, it's not a meaningful difference... ISO 1600 on the D600 D5200 is like ISO 3200 on the D5200 D600. Red's seem slightly stronger on the D5200 (not by much), and blues appear cleaner on the D600. The image samples have visible focus variations and differences in setup, but the broad trend is there: the difference in image quality is one stop. Depending on who you are, you either see that as:

  • Woo Hoo!, FX gives you a whole extra stop of image quality!
  • Wow, FX only gives you an extra stop of image quality!

In the early days when ISO 1600 was the practical limit of DX cameras (never mind the "HI" settings), an extra stop was a big thing. Now that cameras can shoot at ISO 3200 and 6400 for emergencies, the difference doesn't seem so great anymore.


In the early days of FX, people were seeing more than 1 stop's difference between the D3 and D300. I suspect that part of that was because the D3 sensor, being a high-end premium part, has being manufactured to a higher specification than the mass market 12mp sensor in the D300. I'm not sure I'm seeing that difference now... it seems like an "equal" playing field, with the differences coming down to sensor area. This makes sense, as the 24mp FX Sony sensor is still somewhat mass market, and has a broader appeal (and is used in more cameras) than the 36mp part in the D800.

What's more interesting is that even though the Toshiba and Sony sensors are different sizes and come from different companies, the out of camera images have a familial look to them. In other words, when you look at the images, both cameras give output that is "Nikon-like". All told, the D5200  looks like a substantial (if somewhat unheralded) upgrade to the D5100.


  1. Good observation about the ISO performance but it's still early to call IMHO! Photographyblog is quite useless in any type of comparison because of there limited set-up and methodology which itself isn't convincing enough! ePhotozone is better and consistent but still not ideal because lot depends on focusing, metering and lighting condition and could easily produce different result under different set-up. Anyway, I think the ISO 1600/3200 results of D5200 is definitely better than the Sony A57 as downloaded them and did a side by side comparison - the grains are finer and devoid of chroma noise with the Nikon having a huge resolution advantage. But the D5100/7K sensor still manages to produce a cleaner image in both ISO 1600 and ISO 3200. The 16 MP sensor seemed to have better Color rendering too in the higher sensitivities. All my observations are based on a 1:1 viewing and it's obvious that the higher pixel count has its disadvantages in such comparisons! The detail captured with the 24 MP sensor is gorgeous though!

    All in all, D5200 is going to be a very interesting entrant in the mid-range DSLR segment as it seems. All the talks of its sensor being made by Toshiba makes it even more interesting. I'm eagerly waiting for someone to come out with a comprehensive and concluding review (vs. D3200/vs. D5100), may be the folks over at DPR but they seem to be rather busy with their new-found love of the M4/3! Meanwhile the D5100 price is too tempting and I might even get one with a nice 18-105VR at the price of D5200 with 18-55.

    Thanks for the good post :)

  2. From what I've heard from in friends in retail, they're running low on D7000's because of the discounting, and presumably the D5100 as well. Once they run out, that'll probably be it for the ultra-low prices for a while. I haven't gone to the trouble of playing with the files yet, but my feeling is that if you downsize the 24mp D5200 to the 16mp size of the D5100, you will still get a better image.

  3. Cool article about the Nikon D5200 it is always nice to see it compared to other cameras

  4. "ISO 1600 on the D600 is like ISO 3200 on the D5200"

    I'm guessing you reversed the cameras in your wording?

  5. That's for the catch, I did indeed.