The F%^&ing Nikon D800 vs. Canon 5D mkIII Shootout
This was by no means a comprehensive test, but I did try to make it as (i) real-world relevant as possible, (ii) accurate, (iii) idiotic.
For my rambling analysis of the tests, read on. It’s pretty boring though. You’ll probably learn more and be more entertained just watching the video again.
Test 1: Stills Resolution
This was done with the cameras mounted side-by-side. An identical Nikon 50mm prime lens was on both cameras (with an adaptor on the Canon). They were focused in Live View on both cameras wide open, then stopped down to f/8. We zoomed near the center of the frame and had slow-moving objects in the frame. So while not a test chart, I did everything I could to make the image sharp and consistent, and having the only difference be the camera bodies. Both cameras were shot raw, and processed in Lightroom RC 4.1 with no sharpening or noise reduction.
While the Canon “lost” the comparison, I was genuinely impressed with how well it fared. I also like its Live View better. When zooming in to check focus, even though the Canon has less resolution, for some reason it is much clearer. The Nikon get very pixelated in Live View when fully zoomed in. Mind you, this is really just a bit of a nuisance, but one that could matter to people that find themselves using Live View in that manner a lot. I imagine it also speaks to why the Nikon fares worse in ISO and moire in video (more line-skipping, the camera’s potential is very limited in video). That said, given the camera that is sharper in Live View or in the actual resultant photograph, I’ll take the latter. So Nikon gets the slight edge in this one.
Test 2: Rolling Shutter
All the video tests were shot at the highest bit rates possible, with both cameras at “neutral” picture profile and all else equal. Yes, fiddling with in-camera sharpening and picture profiles could have made any of the video tests look vastly different. That was beyond the scope of this experiment, and frankly beyond the scope of what many shooters want to bother with. Let’s see first how they come out of the gate and then go from there.
There isn’t a whole lot to say on rolling shutter. Both cameras are extremely insanely better (speaking non-quantitatively) than their predecessors. There’s still some improvement to be made for extreme-cases of shaky-cam sports shooters or matchmoving fast-moving shots. But it’s so minimal that I’m going to stop worrying about it.
Test 3: Video Resolution
Again the cameras were side-by-side. I didn’t get the alignment as good as I would’ve liked, which is why the scale looks a bit off in the close-up. Keep in mind that’s 400% so it’s exaggerating everything. I did experiment with sharpening the Canon (which I did not end up featuring in the video) and it looked nearly identical to the Nikon in terms of sharpness. It did however have more noise (which the sharpening only accentuated) and less color detail. In the later test we see that the Nikon fares worse with the moire, so I think that’s what we’re seeing here. The canon is just more aggressive about smoothing away potentially-moire-inducing details. It’s like their virtual optical low-pass filter is more aggressive. I’m also very confused as to why the Canon has so more more compression artifacting/noise when at these low ISOs. Both cameras are recording to their highest bit-rate mode, which for Canon is WAY higher than the Nikon. It should be much cleaner; it’s not. In any case, they’re both an embarrassment in terms of resolving 1080 lines of resolution. I call it a ‘lose’ for both with a slightly bigger ‘lose’ for Canon.
Test 4: ISO in Video
Whoah. That’s not what I expected. Holy hell. I did think the Canon would do better, but I expected it to do so by really crushing out the detail. No, it does a pretty amazing job all around. At no point in the higher ISOs am I thinking “yeah but the Nikon is still better on…. this aspect“. The results are so emphatically ass-kicking that I want to go back and double check my work. I was very careful though– I think it’s legit. Which is crazy because in stills it seems like they’re hanging way closer. This does however, raise the question: Would you prefer the camera that is slightly cleaner in the normal ISOs you tend to use (like in the previous resolution test)? Or the one that is better when you get to the insanely high unusable-anyway stuff?
OK maybe that’s my Nikon bias sneaking in.
Note: Higher ISOs were simulated by boosting the gain in post. Since other cameras can shoot higher ISOs, and since we can add gain as needed, I thought it was fair. Plus… sunglasses? Really? That’s not a sanctioned “layer” in stripping rules. So I had to push things a bit farther to make it interesting. The image to the right is my exposure chart for the steps I went through on both cameras.
Test 5: Stress Test
Hearing the crunching sounds as the Canon gets smashed to bits is one of the most satisfying things I’ve ever filmed or been a part of. I will never tire of rewatching it.
Test 6: ???
This was not comprehensive, we didn’t look at:
- How the stills hold up by other metrics, cleanliness, bit depth, dynamic range, etc
- How robust the video codecs were for grading/post processing
- How the in-camera settings can optimize video image
- Sound quality
- Ergonomics/shooting speed/usability
- Tons of other things that people on the Internet will remind me of
We may do more tests going forward. But if I really tried to do a comprehensive analysis, we wouldn’t have gotten it done until the cameras were obsolete.