Last night we put 3 Nikon cameras up against the new Nikon D4 in a studio ISO test and the D3s won. Don't get me wrong, the D4 files look fantastic, and beyond ISO 6400 the D4 did have slightly less color noise compared to the D3s but when we compared the images side by side we saw a pretty significant drop in contrast and sharpness. Even though the D4 has more megapixels, it appeared to be capturing less detail.
We have all seen still life camera tests where the tester will shoot a table full of objects ie: paperclips, crayons, bottles, etc. I personally shoot people for a living so I wanted to do a test with a human face. I called Reese Allen and Steven Hyatt over to help me out with a test. Nobody wanted to model so I sat on a stool for 45 minutes and attempted to remain perfectly still. Our goal was not to create a "pretty" image but rather an image that has details that you would find in the real world. My face was lit from camera left with a 7 foot Octa so that we would have a nice even gradiation across my face. From behind me camera right my cheek was blown out just a little with a large strip box. The background is simple dark grey paper and my shirt has a simple pattern with a little color.
We started shooting at the highest ISO for each camera and went towards the lowest ISO. At the beginning we were forced to shoot with the modeling lights because the ISO settings were too sensitive (even at F22) and once we reached ISO 6400 we switched over to strobing with the Profoto D1 monolights. At this point we left the camera settings the same for each shot and simply raised the flash output one stop as we lowered the ISO for each shot.
We tested the D4, D3s, D3 (same sensor as the D700), and D7000. All of the images were shot at F8 at 70mm and we zoomed out to approximately 46mm on the D7000 to make up for the 1.5 crop sensor. We set all of the cameras to the same picture settings and we shot RAW+Jpeg.
During the shoot Steven was the one actually taking the pictures and due to the D4's new screen, he said the images look significantly better on the back of the camera. Once we got them on the computer that was no longer the case. Images with ISOs up to 400 all looked pretty much the same except that the D4 images were the softest of the full frame cameras. The D7000 was slightly softer than the full frame cameras.
In low ISOs the D3 and D3s images looked identical. The D4 shots have 4 more megapixels so they are larger in size but they are softer and have less contrast.
Once we got to ISO 1,600 the D3s and D4 started to pull away from the pack as the D3 started to become noisy and the D7000 even noisier. At ISO 6,400 we began to see the slightest hint of color noise in the D3s files but the D3s files were still significantly sharper. At ISO 12,800 the D4 was still doing a fantastic job of handling color noise in the shadows but with each ISO stop the images were getting softer and softer. The D3s was introducing slightly more color noise but was remaining sharper.
Could this be a lens issue?
We were all having a hard time believing what we were seeing and we all agreed that this must be a lens issue. The test was shot using 4 different cameras and 4 different 24-70mm lenses. We decided to go back into the studio and shoot 4 shots on the D4 with each of the 24-70mm lenses. Each of these shots looked identical so all of our lenses were performing fine.
After we had finished for the night the only thing that we all agreed could be going on here (besides the camera actually being softer) is that the lens needed to be calibrated in the camera. We noticed that the focus seemed to be falling off a little quicker on the D4 than the D3s or D3 (IE: my ears were more blurry on the D4 than the other cameras). That led us to believe that perhaps the lens was front focusing by a few millimeters. We didn't have time to test that theory
RAW vs Jpeg
At this point we had only been comparing the Jpeg images so we installed Nikon View to look at the RAW files up close. Nikon's software did a fantastic job of making the RAW files look almost identical to the Jpegs so we decided that test wasn't fair. We downloaded Lightroom 4 and opened up a few of the RAW files. To our surprise the D4 files did look significantly sharper. We exported those two files as TIFFs and compared them up close. In the example below I downsampled the D4 files to 12MP so that the images match in size. This is a "fair" comparison for printing from a RAW file.
I've heard many people say that the D4 is "one stop better than the D3s" and that simply is not true. The D4 does do a better job of handling color noise in extreme ISOs but it does this at the price of image sharpness. It appears that Nikon's software is simply "blurring" the color noise to make it blend.
So then you may be wondering, is it worth upgrading from the D3s to D4? Well that all depends on what you are doing with the camera. Steven Hyatt who helped me with this test has owned a D3 and D3s for years and he has a D4 on the way. After this test he said he will be returning the D4. I personally will be keeping the D4 as I will be shooting video on it 50% of the time and I LOVE all of the video features. After this test I see the D4 as a D3s with great video options and I am totally fine with that. The camera also has a ton of other new features like a new focusing system that may be enough to sell many shooters but at this point, if you are only interested in image quality, I don't think it's worth the upgrade. If you don't down a D3s and you are considering buying either a D3s or a D4 I would totally suggest buying the D4 for the additional $800. The D4 will retain it's value since it is a brand new camera and the other camera updates (especially the autofocus and video features) make the $800 a small price to pay for a camera of this caliber.
Keep in mind that if you primarily shoot with a low ISO setting then this is the wrong camera for you. A used D3 or D700 will get you the exact same quality image in the studio as the D4. If you are a studio shooter hold off for a few more weeks and check out our review of the 36mp Nikon D800.
Update: Our good friend Nasim over at Mansurovs.com just did similar, but far more scientific, test comparing the image quality at all ISOs between the D4, D3S, and D3. After looking at his results it does appear that our D4 was having the front focusing issue that I mentioned above. In his test, the D4 does appear to be slightly sharper BUT the conclusion is the same: The D4 is not a significant improvement over the D3S in terms of image quality.