I just did a small comparative set of shots between my D7000 and my D90, two of the Nikon cameras I have at home.
The comparison should be called "informal", since I didn't bring a tripod. I just snapped a few through-focus shots at reasonable exposure levels - at base ISO for each camera. I did bracket exposure +/- 2/3Ev in third steps to get the raw exposure (visavi DN clipping, blown highlights) exactly equal. This meant that the D90 was at ISO200, F5.6 and 1/640s and the D7000 was at ISO100 and 1/320s.
It turns out that with the indicated sign as a white reference, the D90 and the D7000 are as exactly matched regarding real ISO as you can get in real-world shooting. The shots were taken with the same lens (85/1.4, older version) at F5.6. The best target focus shot from the "best" exposure was chosen from each camera and subsequently converted to tiff in an "equal settings" raw-converter, raw therapee. I then checked those conversions against LR3.4, and found that sharpening and other corrections seems to be very equal between the cameras in the Adobe base camera profiles.
Having checked this, I then did the conversions in LR3.4 once more, but with my custom colour profiles to make colour and contrast curves more equal. The LR conversions are the ones shown here.
First, the very mundane scene, a small square/greenery that I often pass when walking the dog. In the overview you can see my: White point reference. In the selected shot this is ~2/3Ev down from clipping in the raw file. I adjusted exposure (in LR) until the sign JUST about clipped, and backed off counting the steps. Equal for both cameras. White balance reference. The other signboard was chosen for both WB settings, I used the WB-picker in the center of those white/gray areas with NR turned up to max - both colour and luma. Then I reset the NR to zero. The chosen image crop part. This is the part you see in the later detail crops.
And straight on to the crops... First with both cameras at native resolution - this is the difference in MP showing. Many have complained about the D7000 missing some "micro-contrast", but to me it seems that the pixel-level sharpness is identical. In earlier comparisons, I've found the D300 and the D90 identical too, both as measured (absolute MTF response with asme lens) and visually.
Then I've scaled the D7000 down to match the 12MP magnification. Here some differences can be made depending on which resampling engine you use - Photoshop's BiCubic is hardly one of the better/sharper out there. I've used a Lanczos-5.
Now, someone might protest that if I sharpened the 12MP image a bit more it would look identical to the downsampled 16MP image. But actually, it won't... You don't have to look very hard to find very clear signs of aliasing and other digital nastyness in the D90 file even at this (equal from the start) quite low sharpening setting. At larger prints/magnifications, this gets visible immediately.
Now on to what most people see as the biggest advantage with the D7000, clean and detailed shadow regions. I took the crop in the shaded innards of the tree foliage and scaled the D7000 to 12MP. A "fill light" of 60 was applied to both.
Here you see a maybe even bigger difference than in the brighter part of the image. Admittedly, a "+60" fill light is quite a lot, but you don't often have the time to set exposure as perfectly aligned to the raw file as I did at this shoot. I usually keep a -1Ev safety margin in exposures to protect highlights (at least when shooting at base ISO). This is the same as "lifting the shadows" in a normal conversion.
Where the D90 - that actually is one of the very best cameras even today, when it comes to retaining shadow detail - has been reduced to a mushy mess where you might suspect what the underlying reality was the D7000 still shows a clean, clearly defined branch grid and even leaf detail.
This was my first try using photobucket.com, hope it turns out allright.
What I wanted to see/show wasn't really the effect of higher resolution, but the effect of the D7000 having a much larger capacity for light input. 16MP times 49K FWC means a total light amount of ->
7.84*10^11 electrons
and for the D90 (that's equal or better than the D300 for static PQ!) : 12MP times 24.5K FWC ->
2.94*10^11 electrons
Thats 1+1/3Ev worth of light input more before the sensor saturates, and this means that you can get even better pixel accuracy out of the D7000 than you do from the D90/D5000/D300 even though it has more pixels in total. You can also sharpen it harder before digital nastyness starts to appear. This goes for base ISO shooting only, at equal exposure the D7000 is only very slightly better than the D90 - until you hit ISO3200, where the D7000 again pulls out a visible lead.
Yes, I do have a few hot pixels. I've taken about 12k clicks with the first D7000 now, and in those the number of hot pixels visible and in any way detrimental to the overall picture quality - zero.
I've seen it when I tested video in very low light, but that was just a test for its' own sake - nothing I would ever use. I guess it would be disturbing if you use the camera as a stand-in video device if you're a wedding photographer.
If you're serious about video - don't buy a Nikon. Don't buy a Canon either. Buy either a dedicated video-camera, or a Pana GH2. It's a lot closer to a "real" full-HD resolution, and it has a lot better picture quality in video. Both Canon and Nikon cheat by up-rezzing a smaller frame (about 1200x1000 pixels) to 1920x1080, and then they call it "full HD". To add insult to injury, the line-skipping and poorly implemented scaling algorithms creates a lot of aliasing and moire.
For 720p in good light, the D7000 works just fine, if you take the video limitations for what they are.
theSuede wrote:
Yes, I do have a few hot pixels. I've taken about 12k clicks with the first D7000 now, and in those the number of hot pixels visible and in any way detrimental to the overall picture quality - zero.
I've seen it when I tested video in very low light, but that was just a test for its' own sake - nothing I would ever use. I guess it would be disturbing if you use the camera as a stand-in video device if you're a wedding photographer.
If you're serious about video - don't buy a Nikon. Don't buy a Canon either. Buy either a dedicated video-camera, or a Pana GH2. It's a lot closer to a "real" full-HD resolution, and it has a lot better picture quality in video. Both Canon and Nikon cheat by up-rezzing a smaller frame (about 1200x1000 pixels) to 1920x1080, and then they call it "full HD". To add insult to injury, the line-skipping and poorly implemented scaling algorithms creates a lot of aliasing and moire.
For 720p in good light, the D7000 works just fine, if you take the video limitations for what they are....Show more →
Thanks for the results Suede. Have you looked into the mirror vibration on the D7000? I did some experiments and found that it was rather noticeable, esp. when compared to the D5100 and 7D. Here are my results:
Nicely performed test. I could do that with D40/90/7000/3x if I ever get around to it... :-)
I can't say that I've tested enough to be able to KNOW that the reason for some of the blurred (less sharp than I would have expected) are caused by mirror vibration.
I do however think that the "vibration" issue is very real, and that it varies between cameras - just as any other mechanical problem, like AF-issues, and some electronical issues (like the banding in some of Canon's newer cameras). Very slight differences from body to body in damping and mirror-box mount tolerances WILL have a large impact on the amount of "mirror-impact" vibration.
Some may have zero problems, other may have noticeable amounts. This is all covered in the manufacturing yield spread, modern economy demands a streamlined production. This means that it's cheaper to run fast and just accept that 5% of the production will be "just a bit worse than we would normally accept". Only about 20% of the unfortunate buyers (20% of 5% = 1% total) will ever complain, and you get a nice production rate hike. Then you balance assembly quality stringency against the number of complaints (guarantee item returns/adjustment requests) you get while in full production mode to estimate how good you need to make the product to hit the sweet spot in profit.
[add]
The D5100 seemed a lot easier to get really "benchmark" sharp pictures with, but the general lack of control-surfaces (different menus, less direct buttons/dials) ruled that camera out for me. I had a loaner for a few weeks. The AF system was also "less than expected", mostly in the AF-C mode. In real-life shooting, the D7000 gives me more keepers - which is the important thing for me. When shooting product or landscape, I always use "Q" mode + remote anyway, so the point is moot.
I very rarely use long exposures, or shoot film in very dark venues (with a DSLR anyway). So - the two biggest faults with the D7000 (that I've found) don't really affect my regular usage.
theSuede wrote:
Nicely performed test. I could do that with D40/90/7000/3x if I ever get around to it... :-)
I can't say that I've tested enough to be able to KNOW that the reason for some of the blurred (less sharp than I would have expected) are caused by mirror vibration.
I do however think that the "vibration" issue is very real, and that it varies between cameras - just as any other mechanical problem, like AF-issues, and some electronical issues (like the banding in some of Canon's newer cameras). Very slight differences from body to body in damping and mirror-box mount tolerances WILL have a large impact on the amount of "mirror-impact" vibration.
Some may have zero problems, other may have noticeable amounts. This is all covered in the manufacturing yield spread, modern economy demands a streamlined production. This means that it's cheaper to run fast and just accept that 5% of the production will be "just a bit worse than we would normally accept". Only about 20% of the unfortunate buyers (20% of 5% = 1% total) will ever complain, and you get a nice production rate hike. Then you balance assembly quality stringency against the number of complaints (guarantee item returns/adjustment requests) you get while in full production mode to estimate how good you need to make the product to hit the sweet spot in profit.
[add]
The D5100 seemed a lot easier to get really "benchmark" sharp pictures with, but the general lack of control-surfaces (different menus, less direct buttons/dials) ruled that camera out for me. I had a loaner for a few weeks. The AF system was also "less than expected", mostly in the AF-C mode. In real-life shooting, the D7000 gives me more keepers - which is the important thing for me. When shooting product or landscape, I always use "Q" mode + remote anyway, so the point is moot.
I very rarely use long exposures, or shoot film in very dark venues (with a DSLR anyway). So - the two biggest faults with the D7000 (that I've found) don't really affect my regular usage....Show more →
I've resorted to using Q-mode as well to avoid the vibration although its slow mirror motion is annoying (desirable to avoid the blur but annoying nonetheless). I believe Chasseur d'images was the first to report about the D7000 mirror vibration and I think they said it's the worst they've seen in quite a while.
Since i find no difference in sharpness between "Q" and a full MUP, there's no need. One press is fewer than two presses, do a few hundred product shots in a day and the difference (in delay/waiting) is obvious.
Not unexpected as the D7000 only provides a theoretical increase in resolution of 14% and a more sophisticated signal processor at the expense of much smaller photosites. Nice to see someone do what you did and actually take pictures with the same lens and then look at the results.
