Thursday, November 4, 2010

D7000 versus D700 versus D3100 in Low Light at High ISO – *** updated 4 Jan 2011

The buzz started as soon as the D3100 and D7000 were announced. This generation of APS-C sensors would pull off the miracle of equivalence with the last generation of FX sensors. The D700 would be equaled by a mere DX camera! Of course, that was wishful thinking for those who thought they would get D700 performance on the cheap.

And the results are in … this new generation of sensors is lovely … the results at high ISO are very useful in small to medium prints. A great 8x10 is easily achievable at 3200 ISO in my opinion … and for some subjects, 6400 ISO should be possible too.

But equivalence in detail retention under difficult lighting is not in the cards.

As has become my habit, I will again demonstrate my point with the excellent images from,  a site that has managed to keep excellent consistency in the ISO series with the mannequin, the only series that really interests me. I consider all of these cameras to be stunning at base ISO, so if you cannot make a great print at base ISO from any of these cameras, then look to your technique and skills. But at high ISO, things get pretty hairy and the camera and sensor make a lot of difference.

All images used here are copyright © 2009-2010 … click on each image to see the large version.

Unfortunately, the D700 mannequin shots appear only in jpeg, while the other two have NEF files, so the D700 would be at a bit of a disadvantage if I proceeded along those lines. Still, the FX sensor should be able to hold its own even with that disadvantage. And in fact, I pile another disadvantage on it by upsizing the D3100 and the D700 to match the size of the F7000. So all crops are 16mp.

So let’s look at 800px images from the three cameras first at 3200 ISO.


D3100INBI03200[2] D7000

D7000INBI03200[1] D700


My analysis is that the D700 already shows a lead in detail retention. The other two are perfectly usable, and kudos to Nikon for that, but the hair has more texture in the D700 shot and the red flowers actually have detail. This, of course, could come from having better picture controls fro the D700 in ACR. But while the subtlety of FX tones are not lost on this comparison, the differences are fairly small at web sizes and score a near-tie for DX. These cameras are already both winners in my book.

Let’s skip 6400 and jump straight to 12800 for this full image comparisons, as I don’t want to fill the article with images when the crops are coming for all three ISOs.







Here, the differences are starting to really pile up. The hair detail is dramatically different, as the D700 retains a lot of texture while the others necessarily clump the hair into an almost helmet-like mass when enough noise reduction is applied to control the luminance noise, which is really severe. Note, however, that the luminance noise is still fairly fine-grained and the chroma noise is almost non-existent, so this generation of sensors is definitely better than the last generation at extreme ISOs. This bodes *very* well for the D400 when it arrives …

A second major difference is the appearance in the DX shots of a blur cast in the shadows and on the red flowers. There is a prominent color shift there, which is a classic effect when the pixels start to get really tiny. I see it all the time on small sensors, and obviously DX sensors are not completely immune at extreme ISOs. I am sure extra processing could handle it somewhat, but once this sort of thing sets in, the jig is up …

So … pretty obvious victory for the D700. In fact, looking at the crops, the D700 looks to be between 1 and 2 stops ahead when you factor in detail and noise, which is right where it should be given the laws of physics and all …

And now the crops to make it all so very clear.

d7000_d700_d3100_crops[1] Reminder – Click through to see the big version!

TANSTAAFL … There ain’t no such thing as a free lunch :-)

Update: Wow … this dog continues to hunt :-) … so I thought I’d do a second set of crops at 3200 ISO, showing different parts of the frame. I processed thought ACR only, and only for white balance, noise and detail. I downsized the APS-C cams to match the D700, the opposite of the method last time. No matter which way I go, the advantage rests with the larger images, in this case the two APS-C cams.

Remember to click through to see the really huge version. And remember to click on that to get full size once in your browser.

My analysis of the crops left to right:

  • D700 easily wins on the hair crop. The APS-C cams are good, but the D700 shows a real difference in actual hair detail over the clumping of the others.Skin texture is also better and I prefer how the D700 handled the red channel, the image is warmer.
  • The D3100 breaks down a tad on the stem where it joins the flower. The edges are suffering from artifacts. Still pretty good in the rest of the crop. The D7000 looks just fine, but I still prefer the more open shadows on the D700 shot.
  • The D7000 loses the fine detail on the label, but retains it on the cork. The D3100 loses even the cork’s fine details. The D700 is the clear detail winner.
  • This one is tricky as there are depth of field and focus point differences here that I cannot account for. What I can say is that the D7000 wins in an absolute sense (notwithstanding the unknown reasons behind the sharpness differences) and that the D7000 shows how well it holds detail in shadows (stitching in the collar’s shadow.) So this is at least a tie between the D700 and D7000.

Even after this update, my bottom line is that the APS-C cameras are improving and are certainly viable low light shooters, but that the universe continues to unfold as it should in that the full frame sensor continues to pull much better detail at higher ISOs.

But the really good news is that the next generation of full frame sensors look extremely promising … I think we’ll be seeing D3s-like performance in all the new cams, assuming Nikon will stop at a reasonable pixel count for the D700 replacement (16mp is just right in my opinion.)

No comments: