OK, time to get your nerd on.
I currently own a Fuji XT-1 and XT-2, several 5d Mark IIIs, and a 5d Mark IV. I’ve owned the d700, edited a ton of d750 files, and spent years shooting with the 5d and 5d Mark Ik bodies.
Here is my real world experience these cameras and sensors and how they all compare, after probably two million shutter clicks shooting portraits, headshots, landscape, adventure stuff, and most importantly (to my business) weddings.
First, a bit of history. After many years shooting Canon, I switched to Nikon in 2011, and then I switched back. I spent more than a year shooting with the D700 bodies (I know — at this point they’re old), SB900 strobes, and most fast primes from 24mm to 135mm and with the f/2.8 zooms from 14mm to 200mm.
Now I’m back shooting Canon with a zoom collection ranging from 11mm to 200mm and primes from 24mm to 135mm. I also use the Fuji system for travel, some personal work, and with the advent of the XT-2, weddings and portraits from time to time. I own the kit lens (it’s quite good!) and the 35, 50 and 85 equivalent primes.
Every winter I spend traveling and shooting adventure sports, and all spring, summer and fall I shoot weddings and portraits. I’ve been doing this full time since 2006, and before that I wrote and photographed for newspapers.
SHADOW DETAIL & RECOVERY
I spend a lot of time shooting in mountain valleys, where I want blue skies as well as detail in the valleys themselves. I may be shooting engagement or bride/groom portrait sessions, or I may be shooting climbers in the shade of a cliff and I want to avoid blowing out the sky.
I hear a lot of people saying that shadow recovery is only necessary if you screw up the exposure, but to be frank, my perfect exposure in these situations looks like this:
Basically, it looks terrible because it’s totally underexposed. But you can see the red highlight alert is just starting to clip the sun dipping over the horizon, which means this exposure is perfect for my use (at least when shooting with a 5d3). Actually, it might be overexposed for a X-series, D750, or 5dIV because the shadow detail is so much better.
After editing, this is what I delivered to clients (with the better sensors, I’d probably want to underexpose so there’s even less blown highlights in and around the sun).
You can’t make an image like this unless you can somehow light an entire cliff band or river valley — which is totally impossible — or push a sensor to its limit by underexposing and then boosting the shadows in post.
So, I shoot these frames in RAW at the camera’s base ISO and expose so I just start to clip detail in the sky, and so the darks in the histogram are just barely to the right of its left edge (at least with Canon sensors — there’s much more detail in highlights in the Fuji sensors). In either case, shadow recovery in these frames is super important.
My experience is that the d750 is the best camera I’ve used in this regard, followed just barely behind by the 5d Mark IV. You can see my comparison of shadow recovery on the 5d III and the d750 here. Bottom line is that 5d3 and D750 are leagues apart.
However, the new 5d4 has really closed the gap. I don’t have one with me to play with, but from fiddling with the shadow recovery on both cameras (different frames and at different times), I’d say that the d750 is only ever so slightly better.
The XT-2 is admirable, but not as good as the 5d Mark IV. The XT-1 is a little bit behind, and the 5d Mark III is maybe 3 stops behind. It’s really quite bad in comparison. I no longer have a 5d Mark II or 5d Mark I, but if I recall, they are both on par with each other, and another stop or two behind the 5d Mark III.
Here is a shot I did on my deck similar to the ones I used for my 5d3 and d750 test.
As you can see, the exposure is similar to the one above with the family on the river bank. The main difference is there is no sun to clip the highlights.
Here are crops of each camera, starting with the 5d3, then the 5d4, then the XT-1, and then the XT-2. I cropped to the same size and then exported to 2048px. So, obviously the 5d4 has been downsampled compared to the XT-1. Sharpening and noise reduction have not been applied to these files.
The Fujis were shot at ISO200, because that’s the base ISO for that sensor. You’ll get the most dynamic range at base ISO.
The Canons were shot at ISO100, so the shutter speeds were different.
Both were shot at the 35mm equivalent with a prime lens at f/5.6, in order to maximize sharpness and have a reasonably deep depth of field. All files were pushed by 1.6 stops in exposure, with +64 contrast, -100 highlights, +100 shadows, and -100 whites. Black points were left untouched.
Again, first the Canon 5d Mark III:
Then the Canon 5d Mark IV:
Then the Fuji XT-1:
And last, the Fuji XT-2:
There is no question to me that the 5d IV is the best of these cameras. It’s also the best Canon sensor I’ve used, hands down. It’s incredible compared to the other sensors in the 5d line. If you aren’t faced with the sorts of scenarios that require shadow recovery, of if you’re shooting in a studio, or if you you prefer to expose for shadows and lose detail in the sky (and many photos like this are quite beautiful) then you probably won’t care.
If you regularly push shadows a LOT, and I do, this matters.
Feel free to download an archive of the above DNG files here to play around with.
HIGH ISO PERFORMANCE
There are a number of things that matter to me about high ISO performance: luminance noise (i.e., grain), color noise (i.e., where you’ll get blotchy gradients that are nearly impossible to fix), and color accuracy/fidelity.
Luminance noise is the least annoying to me as it looks like film grain, and it isn’t as obvious when printed. It’s also the easiest to fix.
Color noise is a pain in the butt to deal with, and can ruin a picture if you can’t get rid of it.
Color accuracy/fidelity is also super important; many bodies, at high ISO, start to degrade skin tones and wind up looking pale or washed out.
Overall, I prefer the high iso performance of the 5d Mark IV. It’s not the noise champ, but the color accuracy/consistency/fidelity/whatever-you-want-to-call-it remains highest or, at least most desirable to my eye, as the ISO increases.
HIGH ISO PERFORMANCE – COLOR FIDELITY
I don’t know if “color fidelity” is the right term, but I’m going to use it describe how colors degrade or change as the ISO gets higher and higher. The 5d IV is fantastic in this regard, and the colors hold up well into the triple digits.
I think the 5d3, XT-1, XT-2 and d750 start to fall apart past 3200, and especially at 6400.
HIGH ISO PERFORMANCE – LUMINANCE NOISE
The d750 is probably the best of the lot, followed closely behind by the (higher megapixel) 5d IV.
I’d say the XT-2 and 5d3 are on par, and the XT-1 is a bit behind.
Luminance noise, to me, is a fairly unimportant metric in sensor performance. It’s relatively easily fixed and prints hold up well if there’s a lot of it.
People who say the XT2 files are as noise free as 5d4 files at high ISO? Well, I don’t know what to make of that. That’s not my experience at all. The shadow recovery at base ISO is comparable, however. Past that, the new Fuji sensors are noisier than the new Canon sensor.
HIGH ISO PERFORMANCE – COLOR NOISE
This matters a lot. It matters when you’re shooting a groom and you find blue, purple and magenta hues in his suit, and they blend together and the suit becomes splotchy. It matters in astrophotography at high ISO, when you see all these hues in the sky and you have huge splotches. It also matters in portrait photography around dusk, where those same hues show up in the sky and you can’t get a smooth gradient from the dark sky at the top of the frame to the relatively lighter sky toward the horizon.
I hate color noise. I actually thought the color noise was so bad in the 5d Mark II (and the skin tones were so red/orange) that I wound up upgrading, then selling one 5d2 body and keeping the other as a backup while I shot with a pair of 5d Mark I bodies. Sure, the camera had a higher ISO capability, but the shots at 3200 and beyond had so much color noise that I found them unusable. So, to me there was no benefit. Both those cameras sucked. That’s when I dumped Canon and switched to Nikon, and the relative lack of color noise on the d700 was a revelation. Compared to the early generation 5d bodies, it simply didn’t exist. The camera did lose a lot of color accuracy past 3200, but the color noise was amazingly well controlled.
The color noise control in the 5d3 was a huge step up from the 5d2. I’d say in my real world use, the XT-1 has better color noise control than the 5d3, the XT-2 is on par with the 5d4, and the d750 is only slightly better.
RESPONSIVENESS AND AUTOFOCUS
Fuji cameras seem to have a mystique about them that I admittedly fell for hard. There is just something about that that is fun to use, and when I use them, I want so much to love everything about them. I tried super hard to look past the terrible aspects of them.
Despite lots of awesomeness, the XT-1 autofocus sucks. It’s slow. It’s slower than the Sony mirrorless bodies, slower than the Olympus mirrorless bodies, and way slower than dSLR bodies. I have shot weddings with this camera and while it can be done, it’s fiddly. The camera does not just get out of the way like other cameras do. So, there, I said it. Tracking is terrible, and one shot AF is pretty sub par. That said, I’ve shot soooo many beautiful images with the XT-1. I really like it. It’s too bad the AF is so bad though. Its other glaring weakness is its general lack of responsiveness (slow to turn on, slow AF response, blackout between frames, etc).
Take, for example, a falling climber (with a rope, nothing dangerous!). With the XT-1, there’s so much blackout between frames that you can’t really track the subject in the air. I imagine it would be the same for birds in flight, for example. I’d have to track them with an open left eye and try to align what I saw through my left eye with what I think I would see with my right eye. The XT-2 is totally different, and feels like a little dSLR.
Next in line is the XT-2. Fuji has made a tremendous improvement with this body. The autofocus is probably as good as a 5d Mark II when used with center points, and it’s better than the 5d II when used on non center points. It’s still nowhere near as good as the 5d Mark III but damn, what an improvement. It’s fast and responsive. There’s no blackout between frames.
I really like the 5d III autofocus, which I think is about as good as the D750 for one shot and tracking. Very very good. I’d be happy if camera tech settled on the AF that’s comparable to these bodies. However, improvements have been made!
The 5d Mark IV has the best autofocus of the lot, particularly in servo mode. When I’m shooting in servo and have the shutter set to focus priority (as in, the camera will try to only take a shot when it thinks the shot is in focus), I tend to get almost as many frames per second as the camera is capable of — and that’s in focus priority! If it’s in release priority, I feel like I get all the frames the camera can deliver in a second. A few fewer are in focus, but the focus priority is so good that I’ll take the small hit in frame rate to guarantee focus accuracy. The 5d III often delivers what I feel is about 50-60 percent of its frames-per-second in focus priority, and it’s wrong about what’s in focus more often.
In one shot, it’s a toss up for me between the Mark III and Mark IV. I prefer to shoot in focus priority, but I found that the focus priority algorithm in the 5d Mark IV is almost too strict. I found it refusing to take shots in poor light (i.e., wedding receptions with no ambient light and relying on the autofocus assist beam from the 600EX-RT strobe) that the 5d3 would have taken (and most of the time would have gotten in focus). So, I had to switch my 1 shot AF settings to release priority, which means the camera focuses more on the press of the shutter button instead of whether or not it thinks the image is in focus. Now, I sort of feel like the camera is faster on the draw than the 5d3, but maybe less accurate. Time will tell how this pans out. I sort of wish there was a middle ground in this priority setting, as I would like to have it balance release and focus priority.
SENSOR COLORS
Perhaps its my lack of experience editing Nikon and Fuji RAW files, but it takes me longer to achieve an accurate skin tone than it does on a Canon. Much longer.
The Canon skin tones are confined to the orange hues while the Nikons and Fujis are “broader” and require tweaking in different tonal ranges in Lightroom’s HSL sliders. I personally dislike this a lot.
BOTTOM LINE
All of these are great cameras, however. Every single one.