A few months ago when I wrote my thoughts on the Canon EOS 1DX, I wondered if 2012 could be the year where Canon makes a serious comeback in the professional photography industry. Some of what I wrote back then:
For the last 4 years I’ve been bashing Canon here on my blog for their poor quality control, poor product releases (50D, 5D Mark II, 7D, 60D) and questionable business practices (5D Mark II, 50mm f/1.2L). I’ve also often mentioned how it seemed like Canon was out of touch with the market and didn’t listen to what professionals were asking for. But with the announcement last year of the EOS 1DX, I think I’m starting to see the long awaited change we’ve been asking for.
Little did I know how much I would be blown away just a few months later when the Canon EOS 5D Mark III was announced along with an entirely new Speedlite system with RF (Radio Frequency) based wireless triggering capabilities. With all the new products in the EOS range that Canon has announced and is still going to announce this year, I really feel excited again to be using Canon equipment. Once again Canon not only has the most complete lineup of lenses, but they also have bodies capable of delivering great images and video. In fact, when you look at the entire EOS range for stills and since recently also for video, and the way in which you can mix these products and use them together, you quickly realize that Canon has an unbeatable ecosystem of professional products for creatives. Nobody else in this industry comes even close.
This review is going to be a long one as I have a lot of things I want to discuss, but for those of you who want the short version, I can tell you just stop reading and go buy the 5D Mark III. You will not be disappointed. In all honesty I can say that the 5D Mark III is currently the best DSLR on the market, apart from the 1DX. It’s a work of art. I bought two myself. And if you’re wondering about the Nikon D800, I can tell you that the 5D Mark III is a much better choice. I wouldn’t buy the D800 even if it was priced at $2000 and the 5D Mark III was still $3500. That is how good the 5D Mark III is. If you think I’m biased, think about the fact that I have recommended the Nikon D700 in the past instead of the 5D Mark II.
Back when the 5D Mark II came out, I was one of the first persons, if not THE first, who pointed out that the autofocus system on that camera was too old and too inadequate for a camera like the 5D Mark II:
With the release of the EOS 5D Mark II camera by Canon recently, I have begun to wonder if the people in charge at Canon’s DSLR division know anything about photography at all. This camera had the potential of becoming a great hit for many different types of photography. But as it is now, it will only be really useful for studio and landscape photography, unless you can put up with its shortcomings.
One of the biggest problems with this camera is its old and outdated autofocus system. It appears Canon just took the 3 year old autofocus system from the old 5D, made some minor changes and put it into the 5D Mark II. The autofocus system has 9 autofocus points, of which ONLY ONE (the center one) is a crosstype point, sensitive at f/2.8. The other 8 surrounding points are less sensitive, being only horizontal-line sensitive at f/5.6.
This is a huge problem for event photographers and journalists who are looking to use this camera, but even studio photographers who want to use autofocus. In low light situations, it’s going to be a real problem using the 8 outer autofocus points to focus on a subject. Many people have complained in the past about the slow autofocus using the outer points on the old 5D, and you would have expected that Canon would have listened and made improvements.
—
So now you have a high resolution sensor, 21 megapixels, but guess what? You can’t easily take pictures that are in focus, eliminating the entire point of having that much resolution available! So now you have a sensor that is extremely capable for low light, high ISO photography, but guess what, you can’t quickly and accurately focus in low light conditions. But hey, you can shoot HD quality video!
When I wrote that, the camera was just announced and not even shipping yet. However, I could already see what was going to happen when it got out and people would start using it. It’s interesting to read the comments again in all my posts about the 5D Mark II, and see how in the beginning I was vastly outnumbered by people who thought I was wrong. But as people started using the camera in the following months, the issues with the autofocus system and soft pictures became painfully evident to a constantly growing number of photographers. The autofocus system on the 5D Mark II became famously known as an “out of focus system,” and along with the 1D Mark III it became somewhat of a running joke in the industry that Canon cameras couldn’t focus. Photographers like Zack Arias and others would frequently make jokes about it in blog posts and during workshops. Even Vincent Laforet, who single handedly launched the video hype around the 5D Mark II with his short film Reverie, would later mention, albeit in a more diplomatic way, that the autofocus system on the camera was old and in desperate need of replacement. Other well known photographers who complained about the 5D Mark II AF system include Lise Gagne, Thom Hogan and Ryan Brenizer. Many photographers who upgraded from the first 5D to the 5D Mark II were disappointed and many of them switched to the Nikon D700. I have documented some of these cases here on my blog. Fake Chuck also reported many of them and some comments from other photographers can be seen here and here. Not surprisingly, the people who were complaining the most about the poor autofocus system and image quality issues on the 5D Mark II were professional photographers who demand a lot from their equipment.
Not long after my initial post on the 5D Mark II autofocus system, Canon Inc.’s Masaya Maeda made an appearance on DPReview where he claimed that Canon couldn’t add a better autofocus system to the 5D Mark II because the body was too small to house a better one. I seem to have been the only one who wrote about this absurd statement, and the online press largely ignored it:
One of the most obvious cases of cluelessness or blatant lying can be seen in this part of the interview:
One of the most common complaints we’ve seen about the 5D Mark II is that it still has the same AF system as the original 5D. Why is this?
“Firstly the market’s evaluation of the 5D’s AF system has been very positive; there have been no complaints from users, with everyone saying it’s very good. Given that, to a certain extent, we think we shouldn’t change it. And also there’s some limitation with size; the AF sensor in the 50D is very big; the one in the 5D is much smaller. If we wanted to have all cross-sensors in the 5D Mark II, it would mean we might have to sacrifice the compactness of the body. It’s all a question of balance of features and benefits.”
This answer is simply a serious insult to our intelligence. Nothing more. It looks like Canon, in all its arrogance, thinks its users are so stupid they can tell them anything they want to. First take note of the fact that the DPReview question started by telling Mr. Maeda that “one of the most common complaints we’ve seen about the 5D Mark II is that it still has the same AF system as the original 5D.” If the 5D Mark II AF system was so good, as Mr. Maeda wants us to believe, then why would people be complaining about it not being improved in the 5D Mark II? This doesn’t make sense, ofcourse, but Mr. Maeda seems to be lacking the intelligence to understand this.
Maybe Mr. Maeda would care to explain to us how it is that the EOS 50D, which is roughly the same size as the EOS 5D Mark II, is still able to hold the “very big” AF sensor which can’t fit in the 5D Mark II. Maybe Mr. Maeda would care to explain to us, how it is that Nikon was able to include a 51-point AF sensor in the D700 an D300 cameras, which are roughly the same size as the EOS 5D Mark II. Maybe Mr. Maeda would care to explain to us, how much more room Canon would need to put an AF sensor like the one in the 50D in the 5D Mark II. 2cm? 3cm? Would it have hurt to make the 5D Mark II body 2cm wider to house a better AF sensor? Perhaps Mr. Maeda would also like to explain how it is that he heard no complaints from users, when the Internet is literally filled with complaints about the old 5D autofocus inaccuracy, especially of the outer focus points. The level of cluelessness and arrogance here reaches astounding levels.
—
I hope Mr. Maeda reads all of these posts and can get in touch with reality soon. What is going to make all of this even worse, is when Canon will be releasing the 5D Mark III in the future, or a similar sized body, with a much better AF sensor, something that is bound to happen. In hindsight, Mr. Maeda’s answer is going to look even more ridiculous.
And indeed today we have a 5D Mark III that is roughly the same size as the 5D Mark II, with a vastly more advanced autofocus system, proving how ridiculous Maeda’s claims were.
I spent countless hours blogging about the 5D Mark II and its issues, as well as issues on other cameras from Canon over the last 5 years. I ran email campaigns and literally spammed Canon executives and media all over the world with my complaints and criticism for years. I even got in trouble with Bryan Carnathan from The Digital Picture which got sorted out after photographer John Harrington intervened, and Phil Askey from DPReview who thought I “clearly have an axe to grind against Canon” and sent his lawyers from Amazon after me.
Years ago I wrote a very angry email to Microsoft CEO Steve Ballmer where I criticized some of Microsoft’s strategies at the time. Much to my surprise he personally replied to me within a day thanking me for my email. I didn’t even expect a reply, much less one thanking me for all the criticism. And I think this reveals one of the important reasons why Microsoft is so successful: They know how to take honest criticism and act on it to improve. And I can say the same thing about some executives at Canon, most notably Chuck Westfall from Canon USA. I’ve criticized Chuck personally in the past, sent some angry emails to him and others at Canon, but to this day Chuck still replies to me when I give feedback and helps me if he can. Everything people say about Chuck’s generosity and knowledge on the Internet is true. And I think it’s because of people like Chuck Westfall at Canon that today we’re seeing Canon making a huge comeback in this industry with products such as the 5D Mark III.
When I look at the recent product launches starting with the 1DX, it amazes me how much of the things we asked for are turning up in the new products. With the 1DX, Canon actually went full frame and scaled back on the megapixels (something many photographers were asking for) and instead focused on speed and image quality. With the 5D Mark III, Canon essentially did the same. Fake Chuck ran a petition with feature requests (actually they were demands) for the 5D Mark III, and it’s remarkable how many of those features turned up in the 5D Mark III.
The sensor resolution remained roughly the same at 22MP, while they included the latest technologies (gapless microlenses etc.) on the sensor and improved the image quality instead. Compared to the 5D Mark II, the 5D Mark III has cleaner looking image quality (about 2 stops less noise), sharper images and the pattern noise and banding issues have been greatly reduced. To be sure, images from the 5D Mark III still suffer from pattern noise and banding mostly in the dark areas, but you only see it when you push the exposure in post. This is still something Canon needs to fix and this has been an issue in all Canon DSLR sensors at least since the EOS 40D. It’s probably an issue that is inherent to Canon’s current CMOS sensor designs. I’m hoping that Canon will find a way to completely eliminate this issue in future sensors and hopefully by the time the next 5D hits the market (or even before that) it won’t be an issue anymore. At Sony they appear to have a better sensor design as evidenced by the sensor used in the Nikon D800, which has no pattern noise and banding issues even while significantly pushing the exposure of an image in post. The noise in the dark areas of an image still looks pleasing to the eye. Canon needs to take notes here from Sony.
With the new DIGIC 5+ processor come features such as in-camera HDR and more importantly in-camera lens correction (chromatic aberration correction etc.). Especially the lens correction features are a huge benefit to the overall image quality you get with the 5D Mark III. I also like the fact that I can set up the camera so that pressing the magnify button while reviewing pictures automatically takes me to the autofocus point that achieved focus at actual size so that I can quickly check if the image is in focus. Speaking of the magnify button, Canon moved the button from the upper right corner to the left of the LCD, and it took some getting used to for me as I would constantly press the wrong button to try to zoom in. But after using the camera for a while I’m now used to the new location. I also like the new rate button, so I can rate an image while reviewing them with a client for later reference. Very very handy.
The 5D Mark III got the same 63-zone dual-layer metering sensor from the 7D which works together with the AF system to make sure the subject that’s in focus is well exposed. However, my experience is that this doesn’t always work reliably. Sometimes images turn out darker than I would expect them to be, and when taking a few shots of the same subject while nothing in the scene changes, you can get a darker image in between. Fortunately the metering system works reliably most of the time (95%). Canon also added the ability to take multiple exposures in a single image and increased the amount of exposure bracketed shots to 7 (up from 3), which is also something I wanted for a long while now.
When I learned that the 5D Mark III would get the same 61-point AF system from the 1DX, I was blown away. I would have been satisfied with the same AF system from the 7D, but Canon went much further by including the AF system from the 1DX on the 5D Mark III. This is a huge improvement on the 5D Mark II and this alone justifies an upgrade if you own a 5D Mark II. This new 61-point AF system with 41 cross-type points is extremely fast and accurate and can even focus down to EV -2 which is equivalent to shooting in the light of the full moon (center point with f/2.8 lens). The autofocus points also cover a large area of the frame, allowing you to frame the exact composition you want at the time of taking the picture. You can even configure the multi-controller to be able to select the autofocus point that you want to use at any time.
What I really love about the AF system is that although it is very complex, Canon has simplified it a lot by including presets for various shooting conditions. You can customize many of the AF functionality to suit your exact needs. Ever since I started using the 5D Mark III, even using a 50mm f/1.2 L lens between f/1.2 and f/2, I hardly get any out of focus shots. Based on my experience so far, I would say that less than 5% of my shots are out of focus. It’s simply incredible. You have to keep in mind that I am very critical when it comes to focus issues, and this was one of the main things I complained about in the past. Now, with the 5D Mark III, I couldn’t be happier.
Not only is the AF system fast, but even the camera reacts much faster with a shutter lag of 59ms (almost the same as 55ms on the 1DX). I will never forget the first time I pressed the shutter button on my 5D Mark III. I gently touched the shutter button and the camera instantly focused and took the shot even before I had a chance to think about it. It’s a world of difference between what I was used to on previous cameras from Canon that I used (40D, 7D, 5D Mark II). I feel so much more in control with the 5D Mark III and feel so much more confident using the camera. Canon also updated the lens microfocus adjustment feature of the AF system, and it is now able to detect the serial number of the attached lens so you can configure settings specific to that lens. So even if you own, for example, two 50mm f/1.2 L lenses, you’ll be able to configure different focus adjustment settings for each one. Previously this wasn’t possible.
The below review by The Camera Store shows how the 5D Mark III autofocus system even outperforms the Nikon D800.
Apart from the fast AF system, the maximum frame rate has also been increased to 6fps (up from 3.9fps on the 5D Mark II). While it’s not bad, I would have preferred at least 8fps. A welcome new feature is the silent shooting mode, where the speed drops to 3fps while the shutter operates very quiet. The shutter durability has also been increased to 150.000 shots. You can now also change the filenames and specify your own custom names on the camera. This custom filename feature also comes with a new firmware update for the EOS 7D and is one of the things Fake Chuck asked for a while ago.
For some reason the 5D Mark III feels so much better and so much more balanced in my hands compared to the 7D. It feels very solid in my hands and comfortable to hold, thanks to the way Canon placed the rubber grips on the camera. I’m not a fan of battery grips as it makes the camera that much bigger and heavier, so I’m happy with the fact that I can comfortably use the camera without one. I have to admit that it still doesn’t come close to, for example, the Nikon D200/D300/D700 etc. in this area.
I’m currently not much of a movie guy myself, although I have been fooling around with the movie recording feature on the camera. So I can’t comment on the movie recording functionality of the camera much. But from the reviews done by others such as Philip Bloom, I am convinced that the 5D Mark III is the better choice for video compared to the Nikon D800 and even the Nikon D4. Check out the below comparison by Philip Bloom.
I also found the below review of the video features very nicely done.
As far as I’m concerned, the 5D Mark III is clearly the better choice compared to the Nikon D800. The images from the 5D Mark III are cleaner with less noise and JPEG images also appear to be sharper compared to the Nikon D800. Even the video quality is better on the 5D Mark III, as can be seen in the reviews I linked to above (for example the one done by Philip Bloom). There’s less noise at high ISO values, and there’s also no moiré and aliasing on the 5D Mark III, where this is an issue on the D800. I don’t know what got into Nikon to go all the way to 36MP on the D800. This is clearly an example of how you shouldn’t be doing something just because you can. Even Nikon users have indicated that they would rather have seen the 16MP sensor from the D4 in the D800, as can be seen from the poll below done by Nikon Rumors a few months ago.
The extra resolution you get going from 22MP to 36MP isn’t worth the issues that come with that jump in resolution. For one, there is a lot more noise in images as can be seen from the screenshot below taken at DPReview.

5D Mark III VS D800 at ISO 6400. Source: DPReview.
You can clearly see that even the 5D Mark II image quality is better than the D800. The difference with the 5D Mark III appears to be almost 2 stops. And the noise on the D800 is an issue at all ISO values except perhaps ISO 100. I’ve seen ISO 320 pictures from the D800 with so much noise I thought it was probably ISO 1600. It’s just awful. In contrast the images from the 5D Mark III are much smoother and cleaner. Here’s another example with JPEG images taken from Camera Labs.

5D Mark III VS D800. Source: Camera Labs. (click image for larger view)
As you can see, again the D800 image quality is worse; there’s more noise while the images also look softer with less detail even at the higher 36MP resolution. This is probably due to the noise reduction that is going on inside the D800 to make the images look better with less noise. But this happens at the expense of image sharpness. This is a clear example of how the extra resolution on the D800 is pointless when there’s also more noise present. You’re going to have to apply noise reduction and this eliminates the advantage you had in resolution making all the extra megapixels pointless in the end, all while adding disadvantages such as the bigger files and slower framerate.
Even on DxOMark, it’s clear that the 5D Mark III images contain less noise compared to the D800, as can be seen from the graph below.
However, when it comes to the overall scores, strangely DxOMark gives the D800 a higher score compared to the 5D Mark III for low light (Sports).
How can it be that the D800 gets a better score for low light high ISO shooting (2853 ISO), while clearly the 5D Mark III has less noise at the high ISO values????? Clearly there’s something seriously wrong here, and DxOMark is misleading everyone with their overall scores. In the noise graph above the 5D Mark III is in the middle, and when it comes to the scores, suddenly the D800 performs better. But even from the sample pictures above, it’s clear that the D800 performs worse at the high ISO values. So how can it get a higher score for low light shooting on DxOMark?? It’s surprising to me that nobody other than Fake Chuck is talking about this.
The autofocus system of the 5D Mark III is also much better compared to the D800. The video review I linked to above from the Camera Store clearly shows this. But this is no surprise considering the fact that the 5D Mark III has the same autofocus system from the 1DX with 41 cross-type points which are spread apart. The D800 has only 15 cross-type points, which are all in the center. This is a significant difference.
I love how the cross-type points are spread out on the 5D Mark III, which is a huge benefit if you want to do off-center compositions and use autofocus, and not having to focus and recompose (with all the disadvantages of that). This was one of the things I complained about as soon as the 5D Mark II came out, and this time with the 5D Mark III, Canon just went all out and gave us the best they have. And it’s awesome!!
Not only does the D800 have an inferior autofocus system, but it’s also slower with a 4fps framerate compared to the 6fps on the 5D Mark III. This difference should not be underestimated if you want to capture the exact moment during an event.
So in summary, I think the 5D Mark III is the better choice compared to the D800 because:
Nikon needs to go back to the drawing board and listen to what people actually want. What the hell were they thinking? I feel sorry for all the D700 users out there for whom the D800 isn’t an option, and the D4 is too expensive. I also don’t think that the coming D600 is going to be a replacement for the D700. If even Ken Rockwell, who is passionate about his Nikon cameras, chooses the 5D Mark III instead, you know Nikon went seriously wrong with the D800:
The Canon EOS 5D Mark III is the world’s best digital SLR. It easily lets us create smooth, clean, beautiful and colorful images in any sort of light. It handles much better my cheaper Nikon D800 and D800E; my 5D Mark III handles and sounds smoother, quieter, feels better and makes better-looking images than my Nikon D800E. The 5D Mark III also has the best LCD ever put in a DSLR, while the LCD on my Nikon D800E is smaller in actual use, and most importantly, the Nikon’s LCD isn’t color-accurate. My 5D Mark III LCD is big, bold, bright, sharp, clear, colorful and accurate, and automatically varies its brightness brilliantly for use in any light. (My D800E also has auto LCD brightness control, but doesn’t work well enough to want to use it.)
…
To shooters upgrading from the Nikon D800 and D800E, the Mark III is greatly improved over the Mark II, while the Nikon D800 and D800E are a step backwards in ergonomics and LCD quality from the D700. Therefore the status quo has reversed; I used to prefer Nikon, while today with the 5D Mark III, I greatly prefer it to my D800E. Pixels impress amateurs, while guys like me who shoot every day for a living are more impressed at how my 5D Mark III lets me get more of the right pixels faster, not just more of the wrong ones.
…
The 5D Mark III is the world’s best DSLR. It’s a huge improvement over the 5D Mark II due to the improvements from automatic color fringe correction, and the greatly improved power, depth-of-field and playback controls. If you shoot Canon, you deserve one of these.
I like how Rockwell remarks that “pixels impress amateurs” while professionals prefer to “get more of the right pixels faster, not just more of the wrong ones.” This is what I have been saying here on my blog for over 4 years now, and it’s what I kept hammering on in all my reviews of the 50D, 7D and 5D Mark II. I want smoother and cleaner image quality, not more megapixels. And let me get that smoother and cleaner image quality much faster and, above all, accurately, so I can make maximum use of the resolution I have available on the sensor. Fortunately Canon listened with the 1DX and 5D Mark III.
While the 5D Mark III is, at the time of this writing, the best DSLR on the planet right now, apart from the 1DX, there’s still room for improvement. Below are some of the things I hope Canon will consider for the 5D Mark IV and other cameras in the future:
Buy the 5D Mark III if it’s in the range of your budget. Seriously, buy it. It is the best there is right now. Unless you can afford a 1DX. In that case, get the 1DX. Like I said at the very beginning, it’s an exciting time to be a Canon shooter. Canon has made big improvements across the board and is coming out with exciting and innovative new products. Also consider the range of EF and EF-S lenses available from Canon, with such gems as the EF 8-15mm f/4 L Fisheye lens and others, and more coming in the near future. Canon’s EOS line of products for both stills and video is unparalleled in this industry. I hope that Canon continues on this new path they have set out and don’t repeat the mistakes they made a few years ago.
The gallery below contains some sample images that I shot with the 5D Mark III. First there are some images I shot with model and paintress Carol Chen Poun Joe, whom I’ve worked with in the past while reviewing the Cactus V5 tranceiver among other projects. During the photoshoot with Carol, I actually used Canon’s new Speedlite system, the 600EX-RT flashes in Gr mode. I will be doing a separate review of the new flash system in the future, but I can already tell you that it freaking rocks!!! I have waited years for Canon to get rid of the ST-E2 and come out with a system based on radio frequency transmission instead of the old optical technology. The ease of use and the way this new system works brought tears to my eyes.
Then there are some other shots that show off the smooth and clean image quality on the 5D Mark III especially in the bokeh areas. Some of those shots are ISO 3200 and ISO 6400 shots and it’s amazing how good the image quality is.
Update July 18th, 2012
I made the comment recently that the 22MP 5D Mark III images upsampled to 36MP using good software would look almost as good as the D800 images with regards to the resolution. It seems like people find it hard to believe this. So I took some images from DPReview here, from the D800 and 5D Mark III both at ISO 100 and put together the comparison below to make it convenient for you. Remember: the 5D Mark III image is upsampled to 36MP. See if you can tell which one of the two columns (Camera A and B) is the 5D Mark III and D800.
The point of this comparison is that the difference in resolution on screen is marginal at best when properly upsampling the 5D Mark III images to 36MP. In print this would be even more marginal. Is this marginal difference in resolution worth the slow 4fps framerate on the D800? Is it worth the overall slower camera and workflow? Is it worth the big diskspace eating files and all the hassle that comes with that (memory consumption, slower PC, backups etc.)? In my opinion it’s not. Also, I’ll take a better autofocus system and better high ISO performance any day instead of this marginal advantage in resolution. The 5D Mark III is the better choice here.
Another thing I should mention is that I also had some trouble with the black focus points on the 5D Mark III while shooting at night. It’s difficult to see which focus point is selected in this case. Many have complained about this issue and Canon is currently working on a firmware update that is going to make this less of an issue.
Update July 19, 2012
I covered a wedding last saturday with both my 5D Mark III’s and after that experience I have some more comments. First, the AF system is amazing. In the over 2000 shots I took with both cameras, using lenses such as the 16-35mm f/2.8 L, 85mm f/1.2 L, 100mm f/2.8 L Macro and 8-15mm f/4 L Fisheye, I can tell you I have yet to find one that is out of focus and where it wasn’t my fault. It’s incredible. With previous cameras that I used, it was possible to get AF confirmation and finding out later the focus was off. Not with the 5D Mark III.
Secondly, the metering system could use improvement. Someone asked in the comments below for spot metering linked to the selected AF point, and this is a very valid request. Using the center point to spot meter is difficult when taking action shots and doing events for off center compositions. There’s a thread on DPReview where this is discussed. During the wedding coverage I had to constantly check the LCD to see if the exposure was correct, or if I had to dial in exposure compensation (usually between 1 to 1.5 stops). This needs improvement. I’m hoping that Canon can release a firmware update where the exposure can be checked based on the selected autofocus point only. Right now this functionality is present in the evaluative metering mode, but this also checks other focus points that have achieved (near) focus:
To work with the iFCL metering sensor, the EOS 7D also features a specific metering algorithm. The EOS 7D always measures focus with all AF points regardless of the selected AF mode. During the exposure reading the EOS 7D looks to see which points, in addition to the selected point, have achieved or almost achieved focus. This information lets the camera know which part of the image is the subject. It then takes metering readings from the zones corresponding to the AF points that have achieved (or almost achieved) focus and combines them with readings from all the other zones. This allows for consistent shot-to-shot exposure, even in complex situations – for example, where there are reflections from a model’s glasses.
Canon could perhaps add a custom function that would enable readings just from the selected autofocus point, which would work as spot metering linked to the autofocus point. Not sure if I’m correct here but this is what it seems like to me.
I’ve also been having some discussions below and I want to thank everyone for their contribution in the comments. With regards to the DxOMark measurements, the 5D Mark III performs better with regards to noise, tonal range and dynamic range above ISO 800. You can check the results yourself on the “Measurements” tab on this link. Below are some screenshots for your convenience.
If the 5D Mark III has lower noise at all ISO values compared to the D800, better tonal range at all ISO values, and better dynamic range above ISO 800, then how can DxOMark give the 5D Mark III a lower score compared to the D800? Especially since these measurements do agree with real world data (stills and video). Yet, DxOMark, in their final score for low light/high ISO performance, give the D800 a higher score. How?
hi Barnett
you need to know how computers does sample down images .
if you downsample an image whit an big amount this can teoreticaly reduce noise up to 100% . i downsample images daily and one of the reasons is that It have a sharpening effect, and it greatly reduces ISO noise.
because:
image noise is always more or less random from pixel to pixel, so the noise in neighbouring pixels is uncorrelated. As you add and average uncorrelated data the average tends to zero or settle to a minimum value. Image data is not random and so there is a strong correlation between neighbouring pixels. Adding and averaging correlated data tends to settle to a representative average. Downsampling is adding and averaging data from neighbouring pixels and producing a single value, thereby reducing the effect of uncorrelated data.
depending on the original difrence in Mp expect a noticeable sharpness increase in a downsampled image compared to an identically sized native resolution.
Downsizing is just a form of noise reduction. When you compare downsized image with converted from raw it’s just the same as compare converted without noise reduction and converted with noise reduction. Of course if you apply strong noise reduction then even files from PS camera can be very clean. At the same time they lost details.
You gave Snapsort a glowing recommendation last year and yet that site ranks NIkons first, second and third and the 5D3 fourth. Their rankings mimic DxO, sorry Fake Chuck.
Thanks for a great review. As a longtime Canon photographer, someone who’s been using a 5D since it first came out, I have come to much the same conclusion about the Mark III. I love it. It’s an extraordinary camera.
But as you point out, there’s still room for improvement. Like you, I have some reservations about image quality. It may be incredible, and it may better than the Mark II, but it still has a long way to go. The banding issues and grid pattern that show up in post-processing are maddening. (And the fact that this problem isn’t showing up in the new Nikon’s leads me to wonder what is going on over at Canon.)
There’s also the slow sync speed, the limited dynamic range, especially in contrasty settings, and the comparatively slow frame-rate. Your review touches on all of those issues.
There are a couple of other factors which you don’t mention in your review, but that I think need to be stated. In terms of everyday use, I consider them critical shortcomings:
1) The spot-metering can not be linked to the focus point. This feature is available on the 1D X and it’s available on all the Nikons. As someone who relies almost exclusively on spot-metering, this is so frustrating. It means I have to lock the exposure if I’m using anything other than the center focus-point (unless I happen to be shooting in manual mode, which is impractical in the field as any photojournalist will tell you). This extra step slows my work down in a big way and is a major distraction when I’m trying to nail the shot.
2) The circuitry connected to the SD card slot is much slower than that of the CF card, meaning that if I have an SD card inserted in the camera, the effective burst-rate slows way down. Several reviewers and bloggers have recently commented on this. The fact that Canon hasn’t acknowledged this problem yet is worrisome to me. It means that the second card is more or less useless in situations demanding lots of shots in rapid succession.
The 5D Mark III may or may not be the best DSLR on the market, I can’t say. It’s certainly the best camera I’ve ever worked with. But it still has a way to go before it’s a perfect camera.
You lost your credibility right here…
“even when you downsize the D800 images to 22MP, they still don’t look as good as the 5D Mark III images. There’s still more noise visible in that case. So I don’t think the D800 sensor is better. In fact, it’s a shame that by downsizing the D800 images to 22MP and giving it an advantage, it still can’t defeat the 5D Mark III image quality.”
And here:
“Also, the 5D Mark III images can be upscaled to 36MP using good software, and you’d hardly see the difference in print. The difference on screen would also be marginal and depend on the shot.”
These comments show a distinct lack of appreciation of the two systems and knowledge of the difference in real resolution for me – I don’t believe you’ve even had the D800 in your hands either.
I haveto agree with Russ Karel,
The upscaling example is kinda pointless. If you can upscale a 22mpx image to look similar to 36mpx image (63%), then surely you can upscale the 36mpx image to look like a 60mpx image and so on. The fact is that there are simply parts of the DPReview test image that the D800 could resolve and the 5D3 was unable to. Uprezing to 36mpx isn’t going to make that detail magically appear.
For example:
1. the globe reveals fine grid lines only visible on the D800/e and Pentax645, while the 5D3 renders the surface with color noise and reveals moire.
http://farm8.staticflickr.com/7267/7602989064_3cf037fdcb_z.jpg
2. The feather detail here
http://farm8.staticflickr.com/7220/7026444741_861a4b62e5_b.jpg
Karel,
A very comprehensive review, but one that appears to have some flaws with regards to the comparison with the D800.
1. The review from the rodeo shootout gave both AF systems a seal of approval. Neither one came out on top. The guest photographer gave the 5D3 the edge on burst rate, buffer and auto white balance, not AF.
2. Images from the 5D Mark III are not cleaner that the D800. Having downloaded raws from the DPREview tests, the raw files tell a very different story in terms of ISO noise. Comparisons show that the 2 are almost identical with regard to high ISO capability with the D800 retaining greater detail at least up to 12,800.
Here’s an example of the raw filse fron both cameras shot at 12,800 ISO with all NR and sharpening disabled. I downsized the D800 to match the 5D3.
http://farm8.staticflickr.com/7251/6870942092_eb3361cc1d_b.jpg
This at ISO6400
http://farm8.staticflickr.com/7050/7016081517_c7bb0dfa91_b.jpg
Can you tell which is which?
3. The Camera Labs link you provided did not compare raw files but OOC JPEGs, and yes, the 5D3 does produce better JPEG. That might be useful for sports and photo jouranlists, but he D800 is not the camra for them anyway.
4. Wjhilethe 5D3 AF is arguably slightly better, the D800 AF has been reported to track better than the 5D3.
5. The claim that the the 5D Mark III has much better image quality for both stills and video is simply absurd. The D800 has far superior dynamic range and all head to head tests have demonstrated this. The D800 does not suffer from any pattern noise. So how can it possibly have better image quality for stills?
6. Wtth reagards to video, the 5D3 is superior at high ISO’s but inferior at low ISOs.
7. The converse of this also means that even if noise of the D800 cropped at 100% looks slighlty more obvious than on the 5D3, a downsized to 22mpx will look the same. In fact, one has the option of applying noise reduction before downsizing and producing a potentially cleaner image with the same amount of detail.
8. You stated a few times that the images look better on the 5D3 but provided no examples to demonstrate this.
Regarding you 2nd point, it looks like the pictures on the right have less noise than the pictures on the left. Which camera is on the right?
Karel, first of all thanks for your review.
You asked “How can it be that the D800 gets a better score for low light high ISO shooting (2853 ISO), while clearly the 5D Mark III has less noise at the high ISO values?????”
Have you tried to understand how DxO calculates the Low Light score? Here is what they say about it:
“Thus low-light ISO is the highest ISO setting for a camera that allows it to achieve an SNR of 30dB while keeping a good dynamic range of 9 EVs and a color depth of 18bits.”
http://www.dxomark.com/index.php/About/Sensor-scores/Use-Case-Scores
So, the low light score isn’t just about noise (SNR), its a mix of SNR+DynamicRange+ColorDepth.
In the end you can have a very clean image, but if the sensor fails to deliver enough DR and CD you get a bad score. In this case, Canon looses points for not having a good ColorDepth, because the Dynamic Range is actually better then Nikon at high ISOs.
Karel, the point you’re making above refers to evaluative metering, not to spot metering. Spot metering on the 5D Mark III uses only the center focus point to evaluate proper exposure. For more on this see: http://bit.ly/MKV1KB.
Dear Karl,
Although I agree with some of what Paulo and Andre highlighted, I support you overall view and enthusiasm for the new 5d. I do believe that nikon could have done a far better sensor had they stayed with a 24 mp resolution and provided better ergonomics and handling.
I do wish to add a few items on the Canon to do list (for 5d mark IV)
1- As mentioned already, the spot metering link to AF.
2- I can do better jpgs from raw than the internal jpeg engine can. So I would think there is room for improvement in how 5d mark iii produces jpeg files.
2- Fix the video quality. We do expect c300 level video quality, not the downgraded crap we get now.
3- auto focusing during video, a bit better than the 650d please
4- Can you please fake dynamic range if you have to? say have an ultra DR mode, where you shoot at iso 100 but the camera really shoots at 400 and does a pixel by pixel hdr with 2 stops expanded DR? Thats the only way I can think of, the 14 bit A/D sensor on the nikon can achieve a 14.4 EV DR. I really want to stop worrying about blown up highlights.
5-Better sensor!, how about the utilization of Canon’s already demonstrated ability to produce a 120 mp experimental sensor in APS-H. http://www.dpreview.com/news/2010/8/24/canon120mpsensor . You can do that 120 mp, then you can do 144 mp in full frame and bin it 2×2 as LRGB true color (non bayer) 36 MP. Well, you can just make it an 89.2 MP full frame sensor with binned pixels, four subpixels , LRGB under one microlens, 16 bit A/D, and since the L subpixel will have no color filter, the LRGB pixel can have good SNR over the whole 16 bit DR. Basically you will get a 22.3 MP non-bayer sensor with 16 EV of DR. I will support Karl in requiring no more than 20-24 mp on 35 mm.
For lenses:
1- Please stop horsing around with lenses. 24-70 f/2.8 with IS is desired and will pay premium.
2- All L primes should have an IS version.
3- The 17-40 f/4 needs new design, and IS(ok my dream) which will be much more useful in quite light.
Thanks
Dear Karel,
Interesting review, it seems that for a lot of your readers only tests are important and they “defend” the pixel King at this moment forgetting that before the introduction Nikon always claimed that 12 mp is more than enough. They all agreed until Nikon came with this pixel beast with 200% more pixels and many changed to the new religion, defending it with all energy. How many are printing their pictures bill board size? Why a photograph is still so attractive for us? Because of the moment captured and not all technical perfection and details, etc. Let’s make photographs and let’s not forget that most of the stunning images of all time are made with camera’s much less advanced than what we have today. Usually you take a decision for a camera manufacture and after having invest in many lenses you don’t change because model a from supplier b has more pixels, better DR, etc. because 1 year later it is supplier b or c with a better model.
Anyhow, after the mark I and Mark II, I do like my Mark III a lot and was able to take pictures I couldn’t take before (flamenco dancers in theater thus fast movements and extreme low light) with a Mark II
Ik vind beide modellen de moeite waard: de Mark III en de zeer aantrekkelijke en getalenteerde jongedame die heel mooi kan schilderen en tekenen. Ik hoop, na aanleiding van je andere post dat je vriendschap duidelijk is
Met vriendelijke groet
as for banding/noise pattern issues appearing when pushing shadows/darks with every single C RAW ever made, it may be the LR/ACR doesn’t read all data as well as DPP or some other RAW convertor does.
here’s some examples:
http://www.flickr.com/photos/mattbicephotography/7133044449/ (underexposed)
http://www.flickr.com/photos/mattbicephotography/7133055985/ (shadows/darks pushed in PP with DPP)
http://www.flickr.com/photos/mattbicephotography/7133254333/ (crop of the same thing above but with LR)
I have been seriously considering upgrading to the Canon 5D mark III from the 5D Mark II and it’s exactly for the reason you mention and after reading your review here, I am now convinced I will take that step. I hesitated because I was afraid it would be the same issue in the Mark III: Concerning the Mark II – Very Bad AF focus system!!!I have had this camera for 9 months and at first it was great, sharp images and focusing was right on and got 80 to 85% good images but after 4 months, I started having focusing problems and it has gotten worse with each use. Lenses were calibrated and did all the usual checks but no matter what I did the AF focus continually gave soft or blurred shots. Didn’t matter if it was AI servo or One Shot, photos came out very soft or out of focus. Even though it would tell me it was focused and took extra time to make sure, the image still came out soft or what I was focused on was blurred or soft but a different area of the image was sharp. It does a lot of back focusing I did a wedding last week and took over 600 pictures, did not even get 10% of the images as good quality, the rest although shot in raw were useless. I have read and talked to several professions about the cause and all say the AF focus has issues, well I can attest to that. I did sooo much better with my 40D and did not have this problem at all and I have the same L lens. This is the 2nd time I have taken it in to Canon for repair and I have to say although I like the camera in a lot of respect it’s useless for me as a photographer if I can’t get the majority of my images to be focused and sharp. Until Canon fixes this issue I would not recommend this camera to anyone. I will be looking very closely at the 5D mark III which their AF system is apparently suppose to be much much better. I will see how it does when I get it back from repair from Canon. For the price I paid for this camera it should have a professional AF system. You can’t use the outside AF points, they are useless and the Center AF point does a very nominal job after a few months of use. And only 9 AF points close together is an issue also. Until they address this issue, I would not recommend the 5D Mark II.
Thanks Karel for your review.
I found your review of the 5d mark 2 very similar to my experiences over the last few years and generally have had a very frustrating time with the mark 2 version and had work with its flaws in mind, in particular as a wedding photographer the couple coming down the aisle, first dance all the usual suspects in low light.
Over the past few months have trawled the web for any negative about the mark 3 and may as well stop now and take the plunge.
Generally agree with everything Yvonne mentioned above and yes, had more success overall with my old 40d.
The only review that’s worth reading is at DX0Mark, there you will get an unbiased review. A comparison of the 5D Mk3 and the D800 clearly puts the D800 way ahead.
The D800e is also rated as the best sensor, D800 next, D600 3rd, 5dM3 way down the list at 13th.
http://www.dxomark.com/index.php/Cameras/Camera-Sensor-Ratings
Portrait colour depth .. 5dmk3 is 15th .. Nikons entry level D3200 is rated better, lol.
I guess I am half expecting for my post to be either deleted or the Canon fanboys to moan that DX0Mark are way off the mark ..
Karel, I’m curious with your experience with 5Dmk3 and AF assist beam with various flashes.
With my 5Dc and 550EX I had no trouble.. just once time I used 430EX and AF kept hunting until I gave up on it but since it was only one gig and 430 wasn’t mine I didn’t pay further attention until I saw this thread:
http://www.canonrumors.com/forum/index.php?topic=10275.0
I own the D800 and two D3s Camera’s having switched from Canon after years of frustration. I do not question your review at all and am sure the 5D3 is a better camera in the real world than the D800.
I am a full time wedding photographer. I cannot use the D800 for weddings, why? like you said too much noise even at the lower iso scale. Raw files are too big which makes processing a nightmare.
One thing I do disagree on is the D800 does have banding issues. However I didn’t buy this camera for weddings as I have two D3s’s. I bought it for the studio and there it really does shine. I’m positive the 5d3 does too however the detail is staggering and since I’m shooting 100 iso I have no issues with that.
I think people these days expect ‘all in one’ camera’s but seriously the true photographer would know the D800 is going to be targeted at landscape and studio photographers. The 5d3 clearly is a all in one camera that does most things well. Lets face it most wedding ‘togs’ either use D4′s or 5d3′s I’d really like to know how many serious wedding photographers out there use a D800. How anyone who does 3 weddings a weekend at say 1500 shots per wedding could edit these files is beyond me. Oh one more thing the DOF on the D800 is so small its incredibly hard to master in a fast paced situation. Do yourself a favor and buy a 5d3 That Af system is worth it alone. Unless of course you have the money for a D4 or a 1Dx
Great comment Mark,
I think the amazement at Nikon need their heads examined. Why they refuse to stick an D600 sensor in a D800 body and create a true D700 successor s beyond me. The D600 is a nice option, but it is too hampered.
Surely they must have figured out that not all D700 owners are upgrading to the D800 because the D800 clearly does not serve the same purpose; thus they can’t seriously believe D800 sales would suffer if there was such a camera.
I’ll keep the D800 for studio etc, but Nikon are are pushing me to invest in a 5D3 and lenses because there is no real D700 successor. It is infuriating.
It just seems to me that two guys bragging about their cars. First one owns a Ferrari & the second one owns a Lamborghini.
Both are high end professional SLRs yet each one shine in some areas, no all. Therefore, you need to choose the one that is suitable for your own use & brings out the best in your shots.
As a wedding photographer, the 5d3′s quiet shutter mode is worth having over the d800. The d800 however can spot meter at any chosen focus point, the 5d3 can’t.
the 5d3 has more cross type focus points…the d800 has a much better dpad than the fiddly controller on the 5d3.
The higher you go in resolution the more pixels their are, therefore shooting at 1.4 on the D800 36mp creates the impression of a smaller DOF if you printed both the 22mp 5d3 and the 36mp D800 to 20×30 the D800 will look much better bokeh wise because it can be printed to a much larger size. This can be a good and bad thing IMHO as I do tend to stop it down a bit more to get the same effect I’m looking for. So that’s good however if you do want to shoot wide open you have to be very good and accurate.
Good article, Karel; I find a lot I can agree with here, especially the emphasis on overall image quality. And I am an amateur–so not all of us want Mp’s. I would really like to see Canon replace the cheap black foam up by the focus screen that dampens mirror rebound. The stupid stuff sheds black flecks into everything like a sonofagun. Pentaprism, sensor, everywhere. What good is it to invest millions in product development and not spend a couple man months finding a replacement for this?
Good write up Karel on what is a great camera, it’s blown me away in the improvements over the mk2!
Thanks for the review, but it is obvious you are a Canon shooter and very very biased. You only inserted clips into your review that attempt to boost the 5Dm3 and omit the clips where the D800 beat out the 5Dm3. Let your readers judge for themselves, here is part 1 of the video you conveniently left out.
http://www.youtube.com/watch?v=omTo7UxbJX8
Here is a vid comparing the D800 to the Hasselblad H4D-40
http://www.youtube.com/watch?v=9UBTE4xpvpk
D800 vs D800E
http://www.youtube.com/watch?v=I3M0VHBFiek
The 5Dm3 is a phenomenal camera as is the Nikon D800(E), know your shooting style and pick your poison, the end!
it is all about AF AF AF AF you can have 80 mp ( Hassie) pixels bud if the focus is not spot on the picture will be trash
so Canon 1Dx and 5D3 about 20 mp make Nikon loose nowedays altouch the body is more solid and qualitycontrol either.
It you like 5D3 not enough go to Phase One bud you must spent 20.000 as a start
kind regards
I’m just reading your review now, months after it was first published. Great job, and I actually find it unbiased and objective despite what readers’ comments indicate. You simply appreciate the 5D Mark III over Nikon D800, so do 90% of the professional I know who shoot fashion and beauty for a living for Vogue and other high-end magazines, all of whom now owns D5 Mark III’s, as primary camera -This is no coincidence.
I have always been an avid Nikon user but last week I requested to sell off my Nikon gear and tomorrow, Monday, I will order a 5D Mark III and a couple of prime lenses, one of them them naturally the 50mm 1,2.
A final comment, I find the territorial pissings of Canon vs. Nikon fans quite ridiculous – People appreciate the history of each brand, and their product-range for their own reasons and needs. Why can’t we just accept that we all have different taste and demands. I have never heard professional photographers argue with each other IRL which system is better. One photographer bring a Nikon to the studio, the next a Canon. No questions asked, except holding and weighing the other system in their hands, both giving an appreciative nod to the other when handing the camera back.
Keep up the good work
You simply appreciate the 5D Mark III over Nikon D800, so do 90% of the professional I know who shoot fashion and beauty for a living for Vogue and other high-end magazines
That’s as much due to legacy reasons as any other. Investing in systems in pricey so there’s little incentive to change other than for a very good reason. After all, professionals who shoot fashion and beauty for a living for Vogue and other high-end magazines would do most of their work on Medium Format anyway.
I just saw this article.
Guy, if you want to read correctly DxO measurements, you have to stay in “print” mode. “screen” mode is just pixel peeping, it makes no sense to look at an image at 100%. An image is meant to be seen entirely, and in this aspect, the D800 beat the shit out of a 5DIII with his last century sensor technology. 180nm vs 500nm process and on-sensor conversion, the 5DIII has no chance.
Please Canon, move up your ass to produce sensors without banding or high read noise.
ahhh dear karel you can’t dare affirming anything because fake chuck said something. this is true! he does the worst review ever, 1000% pro canon, as you are, he uses all issues he can find about nikon to prove you bought the good stuff but never talks about the ones the 5D3 is outperformed.
I tell you, i am D800 user now, i was 5D2 before. I am absolutely not a faithfull brand guy, i just want the right thing for my money so i give you my most objective observations:
First of all, i tried all cameras. the 5D3 is the most delightful camera to use i have to admit. AF is snappier and accurate like a tamed dog. menus are as always more clear, the lcd is better and i personnally prefer the canon colors. jpegs out of camera are better (i don’t care i shoot raws). the handling seems better to me, a surprising improvement compared to the mark II i owned. just would love some more high to fit my little finger. but as said, the color noise and banding effect at 100 isos is just unbelievable, unconceivable.
The D800 shows somehow A BIT more noise, but after post processing the raw files match the same level as 5d3. the wide dynamic range is a huge improvement! you don’t have to choose anymore between highlights or shadows. you can recover that easily dark shadows or highlights where the canon can’t. The AF grab target in extremely low light situations and you can have a VERY useful AF illuminator in total darkness coupled with AF red light in the viewfinder. Also if you read some more serious review in some rare occasion 5D3 can show moiré when D800 doesn’t. seems weird but true! the nikon does also show some in some occasion.
now if you really want to spit on nikon, as i told you i’m not a faithfull customer, i give you some real reasons. but there are none directly about image quality.
first of all, the D800 suffers from minor bugs and lonngggg image treatment especially using the live view. Digit 5+ processor is faster than Expeed nikon. the battery life is better on the canon. if you take time to use the lcd screen the autonomy falls from 750 to 450 images! this is ridi-nikon-culous. while my 5d2 handled close to 850 in the same conditions. it is said also the case for new canon products but not as bad and i heard for such new norm in japan, but i guess it is to sell more batteries! the new AF system should indeed be tested by dxo, who i presume are using live view mode which is perfectly accurate on every camera. lots of d800 and certainly d4 from the first series (like mine..) suffers from the same left focus issue, and in my case a terrrible backfocus in the center issue making lenses unusable under 2.8 aperture or at the risk to miss 50% of your shots with wide open primes. the AA filter seems to me sometimes a bit too strong, making some fine details not as sharp as you would get unless you close your lens at 4-5.6. yes… they had to do that to justify the existence of a D800E after all!
Now you have real stuffs about nikon. and i don’t even mention the fact that my camera came back from repair with left focus issue fixed but with a damn front focus in the center now… -_-’ but again about IQ itself, when you can get your shots accurate, canon and their 5d3 -as we say in french- “went to pick up strawberries” while nikon and sony made a piece of art with this 36 Mpix sensor.
That said, if megapixels doesn’t count, dynamic range does. a lot! and i’m really looking forward to see my camera fixed. if i’m still not 120% satisfied of it i will just have to consider selling it as long as its price is high. and at this moment will looking maybe forward to the so rumored canon 5Dx 39-46 Mpix with a non dissimulated pleasure.
Also i would like to remember you not to forget Nikon might be not your real ennemy on a long time term. fuji and its Xtrans sensor from the Xpro one are from one part the BEST in term of reproducing skin tones under everykind of light, but also capable to match if not outperform iso noise even at 12800 from CANIKON on a ridiculous APS-C sensor! of course, without AA filter and moiré free
and this, on a 1500 euros camera. which is btw in the same category as a 5000 euros leica M9, precisely outperformed in everything. ah.. leicaists are bying leica because it’s leica i forgot.. since rumors are talking about a X-pro2 full frame, there is something to be worried about your favorite product, then you will have to accept the superiority of another brand.
Well said Thomas,
I have both cameras, so I think I can offer the most comprehensive comparison between the two.
First of all, they are both lovely cameras, and to suggest one is clearly better than the other is frankly baseless. I hung on to both because I can’t bear to let either one go – an expensive proposition no less, as I have a significant investment in glass for both systems.
What I have come to realize is that both systems are a compromise on some level. Since Nikon upgraded the firmware, the differences in AF performance have closed significantly. It now really comes down to glass. In a studio test, with both mounted with their respective 70-200′s, the D800 was superior, but the 5D3 performed perfectly.
When comparing both with the latest 24-70s, the Canon LII lens is superior until F5. At that point, the D800 pulls away.
Generally speaking I have to be more careful with AF on the D800. It is still not as good as the Canon AF system and the 36 mpx resolution is very unforgiving on AF accuracy and motion blur.
On the 5D3, I have to really cautious about exposure. One can get pretty lazy on the D800, given the latitude one has to fix problems in highlights and shadows. Not so on the 5D3. In fact, when it comes to back lit subjects – especially blonde people or bald, the D800 smokes the 5D3. The clipped highlights on the Canon files are beyond salvation. The D800 copes really well in general.
As for pushing shadows, well, we already know that score. So
The 5D3 is definitely a better generalist camera and I tend to use it more often – mainly out of practicality. The raw files are obviously smaller and more manageable so it’s better suited to every day needs.
On the other hand, the D800 files always give me a buzz when I open them for the first time. I don’t get this rush with the Canon files. The Canon cannot touch a D800 file that is perfectly sharp. I can spend ages inspecting a landscape shot looking at all the details – faces on people I didn’t even know where there.
Karel, this is a great review except for one thing. People are not going to take you seriously if you criticize the DxO results without taking the time to understand it first.
DxO does their comparisons at an equalized print size, as it should be done. When you do that, the D800 images show less noise. But if you insist on comparing cameras at pixel level (which is completely meaningless), you will arrive at the incorrect conclusion that the 5DIII show less noise than the D800.
The only real disadvantage of the very high MP of the D800 is the increased file size. But if ultimate image quality is what you are after, the D800 is the better camera.
Having said that, image quality means nothing if the image is not sharply focused. So I hope that people like DxO will start doing AF performance testing as well in future.