Nikon D800 or EOS 5D Mark III

[RRRoger]

you shot THAT at ISO 12800? QUOTE]

No way to be sure as there is no EXIF on movie frames.
The ISO kept MAXing out when I panned into the shadows.
It is set for Auto ISO 12800
However, I think that image frame was only 6400

 

[Danielvilliers]

Danielvilliers convinced me that the 5D3 is not actually better than the D800 for low light: the 5D3 looks a lot cleaner, but it is also a lot softer out of camera; you can apply tons of noise reduction to the D800 before you lose as much detail as the 5D3 has already thrown away by default; and after that it's not clear at all that the 5D3 is the winner - if only the D800 was aliasing/moire free... - also, I like Philip Bloom's reviews, but you have to take them for what they are: he's talking about how the camera feels to him, and how much he likes shoting with it, so it's an extremely subjective thing, and leaves many areas completely untouched

I come from photography and I think that it is gaining a certain maturity in terms of test. Dxomark has set a certain standard and methodology that is very good. The only thing is there general note which can be a little bit subjective. But if you look at each marks in terms of sharpness, Iso and DR it is very well done. One thing for example that was not tested well before is output resolution. Cointrary to everyone that was comparing at pixel level for noise, DXO standardize on 8 megapixel because in the end when you have to compare two images you will compare them at the same size in terms of print or screen resolution on a monitor.

If tomorrow you do a lowlight photo for a wedding and you have to print it in an album and a photobook. You won't print it 3 times the size because you have shot on a 36 megapixel d800 compared to a 12 megapixel d700. What happens is that you will have to reduce the size of the image for the d800 compared to the d700 and this causes a rounding off that will decrease the noise. So there was a big debate because many where saying that the 36 megapixel d800 would be to the detriment in high ISO. The fact was that at pixel level, indeed the d700 was better, but when you resized it at 12 megapixel it was about a stop better at the higher ISO.

Unfortunately in video you don't have these high quality test. Everyone can sell a camera and say that it is 1080p even if it is barely 720p like the Canon 5d3. The 1080p is more like a container until now where everyone can put anything. Now it is up to me but from the Nofilmschool test I see them closer than many thinks at first sight. You have to take into account the difference in resolution, one stop exposure difference in ISO and contrast (burn shadows) to have a better idea between the two. As his last test shows with neat video is that it is closer than we think, as the better details affords you better NR.

From my experience the same can be done with the hacked gh2 like here https://vimeo.com/32204670 . The extra detail lets you have NR to such an extend that you can get good resolution at a certain resolution. Would you say that the 5d3 example has more detail than the gh2 example at this ISO. From what I have seen I would say no, so in the end they are comparable. Now it would have been better to have them both side by side in a proper test. But this was just to show that ISO level also correlates to detail/resolution for a good comparison.

 

[Danielvilliers]

This ISO difference has been noticed from as far as the first 5DvsD200. I also noticed when comparing the 5dmkII vs d7000. Always a 1stop difference.
The 5DmkIII however, is cleaber than the d800, the nikon is not that good passed 1600 in my opinion.
And about the philip bloom test... Well I still don't understand how people can take those reviews seriously. Testing cameras while having beers in a bar???

Yes, I am with you on that, I can't understand how people can judge from these kind of test with so much variables. So if at one time you had 10 people around him (like his bts shows) and at any moment, 5 of them lose their curiosity and decide to go back to the bar to have another beer so twice more light falls on the subject!!!!!!!! Where is any objectivity. But don't say it too loud because there are people that take it as gospel.

 

[Danielvilliers]

After seeing several comparative tests on the internet, I have also asked Philip Bloom if he did notice a difference in the calibration of ISOs between the two cameras. He said no, and told me the 5dIII was easily better than the D800 in low light.
Saturday night I had the chance to test the mark III against my partner's D800.
My conclusion confirmed what I had already seen on several tests:
ISOs are rated differently between the two cameras. The D800 is definitely 1 stop (or slightly more) brighter than the MarkIII for the same given ISO setting, aperture, and shutter speed.
As a result, ISO 100 on the D800 looks like ISO 200 on the markIII, and ISO 1600 on the D800 looks like ISO3200 on the MarkIII. The effect is more pronounced at lower ISO settings.
Of course, you get nothing for free, so brighter means also noisier, but this is just normal.
My conclusion is that saying that "the MarkIII being by far better than the D800 in low light" is just a marketing gimmick, as ISOs are rated differently. Of course the D800 at ISO 3200 has more noise than the Canon, but then again it is as bright as ISO6400 of the Canon... The D800 also appears to have a bit more dynamic range.

Just a heads up for potential buyers.

Thanks for your honest testimony, so many people judge only by the brand that they have. I am a Nikon shooter (photographer) that bought a 7d because Nikon had nothing in the video comparable 2 years ago and a gh2 afterward because I think it is damn good camera (more so the hacked version). So I try to be impartial and not some dumb fanboy because it does not make me advance in any way. I have saw what you have said above and just wanted to tell people as you did from my observation that the difference are closer to what some people are saying.

 

[Samuel H]

* I have always made a point of owning a set of lenses that could work on anything. Right now, they are mostly vintage Leitz for Leica-R. I sold a couple of M42 lenses because they wouldn't work on Nikon, even if my camera at the time was a Canon (I also had great hopes for the replacements). If tomorrow the killer camera is an Olympus, I'll go for that, even though I've never touched an Olympus camera in my life.

* I understand the concept of normalizing for comparison. It was a long, long, long time ago that I wrote this half-botched post: http://www.similaar.com/foto/mpix/mpix.html (yes, the big-sensor pictures are out of focus; it was one of my first inroads into lens/camera testing; but it proves that, after downsampling, a 333 Mpix fullframe sensor wouldn't necessarily show more noise than a low-mpix one) (I'm feeling nostalgic, I may redo these tests properly...)

 

[Northainan]

I disagree, it really boils down to what type of still shooting you do. If you do fast burst sequences for sports, redcarpet events, ect... the Nikon has a major flaw, it takes the camera 2 minutes, yes no mistake, it you burst shoot, it will need a full two minutes to buffer and save the photos, and while buffering, the camera is non responsive, NOTHING WORKS until finished. So anyone that buys this cam for stills should keep that in mind. But in 2012 to have a cam take two minutes to buffer is silly. The Canon slows down but never stops taking pics, and thats the difference between the money shot and NO shot.

So Nikon engineers still have not solve that problem? I have the Nikon D2X,and I shoot basketball sports. And it does stop shooting when I am in continuous shooting mode. So I shot games with the much cheaper Canon 20D and it nails just about the majority of the continuous shots while auto focusing at the same time. But I think that's an issue with the Nikon's flagship D4 though.

 

[RRRoger]

So Nikon engineers still have not solve that problem? I have the Nikon D2X,and I shoot basketball sports. And it does stop shooting when I am in continuous shooting mode. So I shot games with the much cheaper Canon 20D and it nails just about the majority of the continuous shots while auto focusing at the same time. But I think that's an issue with the Nikon's flagship D4 though. QUOTE]

You Canon Trolls must have used the slowest and/or uncompatable memory cards in the Nikon Cameras you are complaining about.
If you want slow inaccurate AutoFocus try a 5D2.
I've never had any of my Nikon's stop shooting for two minutes except when I released the shutter button.
I've shot the D1-4 (including the D2x) + D200, D300, D90, D3100, D5100, and D800.
When the buffer is full and you have a fast memory card, the stills slow down to about 1 per second or much faster with the D4 & XQD card.

I personally set my cameras to 4fps, but I have sent Canon shooters home after they crowded in next to me at an Event.
All I had to do was change my setting to CH and hold down the shutter on my D3.
After a long burst, they would just turn tail and disappear.

As for the D800, it totally blows away the 5D2 and I will keep it over the 5D3.
My Videos have been very good and the stills stellar.
I filled my buffer yesterday in a low light shoot under the Redwoods and the card and camera were still fast enough to get all the runners.