It'll be a bit longer before I can finish my own controlled tests, but all the D4 and D800 samples now starting to appear on the Web allows us to make some preliminary observations.
I believe I wrote this before, but I'll repeat it: the D4 images look a lot like D3s images. The D4 has some dynamic range benefits at the lowest ISOs, while the D3s still appears to create the HI ISO values slightly better. The operative word for the most part is "slight" (Base ISO dynamic range may be the exception). In my opinion, not enough to make a tangible difference to decision making.
Image quality-wise a D4 is visually an awful lot like a D3s with 16mp instead of 12mp, and more dynamic range at base ISO. That's a very nice jump in sensor capability considering that the D3s was essentially state of the art. But it doesn't truly dethrone the D3s as a low-light camera. Technically, you could downsize the D4's 16mp to 12mp and get a slight benefit, but in practice I'll bet that most D4 users will just consider the change in pixel count a "cropping flexibility."
I believe some slight differences in channel response between the D3s and D4 exist, as I'm seeing higher blue channel WB shifts on the D4, and thus different light types could trigger slightly different response. But that difference, too, is small, and I'll need more time with the D4 in my hand to figure out how to describe this. Overall, the D4 is turning out to be more impressive than most of us were guessing it would be given the pixel count boost. Good job, Nikon.
That said, the D3s doesn't immediately go into the garbage bin. It still is an amazing camera, just 12mp. If you can pick one up cheap (doubtful, but there's always those who are quick to abandon good for good+), it'll still take great images tomorrow. To put it a different way: if someone were to invite me to photograph the Olympics this year I wouldn't worry if all I had was a D3s.
Meanwhile, the D800 is a slightly different story. To date, most of the discussion has been about D700 versus D800, but I think we now have to compare D4 versus D800. Here, things are going to settle out just about as they did with the D3x and D3s, I think. If you downsize a D800 image to D4 size, the D4 image is visibly better at some ISO point. Again, I need much more time trying to quantify and explain the differences, but they are clearly apparent to me.
So let's rewind back to the D3 era. What I wrote about the D3/D3x combo boiled down to this kind of gross simplification:
- With the D3 I'd work with a complete disregard towards ISO up to about ISO 800. It's a camera that you can set to Auto ISO 800 without seeing any high level visibility in image differences.
- With the D3x, I always wanted to work at base ISO, and always pulled the ISO down as low as the situation allowed. Any bump in D3x ISO basically undercut dynamic range and increased noise production, and it was marginally visible at any 2x increase in ISO, and definitely visible at a 4x increase.
- Thus, I tended to use my D3x only up to ISO 800, at which point I always felt that it was better to use the D3 (and especially D3s).
So here we are with a D800 and D4. Did anything change? So far, I'm betting no. Substitute D4 for D3/D3s in the above, and D800 for D3x, and I'll bet I'll end up writing something very similar to those three bullet points.
Oh, I might change that third bullet to ISO 1600 (or 1000 or 3200 or whatever the number turns out to be), but so far in every sample I've seen, whether it be from Nikon, from a pro photographer I know, from a Web test sample, or the few samples I've been able to take myself under uncontrolled conditions, I keep seeing the same thing. (Before anyone says "well, you're seeing what you expected to see," no, that's not the case at all. If you recall, I wrote not too long ago that I didn't think that Nikon could match the D3s while increasing pixel count. The D3 was state-of-the-art when it appeared, the D3s trumped that by almost a stop, now the D4 moves beyond that mark yet again. That's one hell of a sensor parade, and wasn't what I expected.)
Now let's get to the title of this little article: "What People See." Here's where things get noisy (pardon the pun). Some of us have been trained in low-level image analysis. Most people posting on Internet fora are not. I see mostly hyperbole and exaggeration ("OMG the D800 is as good as a D4"), as well as a few pixel peeping party poopers ("OMG look at all that noise at the pixel level").
Comparing image samples is not an easy thing to do. Looking at JPEGs on the Web created from converted raws of not tightly controlled captures is not conducive to making high quality analysis. I can make many cameras "look like" other cameras if you give me that many loose variables. It's one reason why I went more to a "read what I write" versus "here's some samples or numbers" approach in my reviews.
To be truly equal, cameras would have to have the same resolution, same dynamic range, same noise propagation, do so under a wide range of test conditions and lens settings, and be post processed exactly the same. I currently do see equality between the D4 and D800 at the higher ISO values.
Just like a D3x had an advantage over the D3 at low ISO values when downsampled to 12mp, so will the D800 have an advantage to the D4, I think (again, very early analysis; I reserve the right to change my mind after a more thorough examination). At some higher ISO value, there will be a crossover point where the D4 image will be better than the downsampled D800 one. I've seen people speculating that this could be really really high, like ISO 12,800, but when I look at their samples, I don't see what they see (plus the samples themselves have issues). Moreover, be careful about what you're looking at. The same test at f/2.8 might show something slightly different than a diffraction impacted f/16. That's why some of us have to spend days running a complicated series of tests before we come up for air with an answer. Until you see the white smoke coming up from the chimney of my office, I haven't elected a new leader yet.
There's no doubt in my mind at this point that the D4 and D800 are going to be great cameras to use. Compared with the D3 and D700 that kicked off the last generation, the bar has been clearly and tangibly raised. Just be careful of people who aren't seeing differences between a D4 and D800, or are stating that a D4 is vastly superior to a D3s, or are making other claims that aren't quite supported by the evidence to date.
The news is good (the D4 and D800 are excellent). Just not necessarily as good as some people are writing in haste. Heck, they may be better. I don't know for sure yet, other than to say that Nikon is continuing to make fantastic progress tuning their sensors and getting results that make for great photographic images in a wide range of conditions. It's going to be fun pressing the D4 and D800 as far as they can go. Indeed, they're good enough that I think they'll challenge my photographic skill.