[Video] Canon 5DM3, 5DM2, and Nikon D800 Low-Light Video Test

Who's got who beat is a big deal for those who are debating where to put their next $3000 -- and rightly so. So if you're interested in shooting video on any of the newest and hotly debated DSLRs out there, check out this low-light, high-ISO video test. The winner is crystal clear -- literally.

SPOILER ALERT: So if that 5D Mark III doesn't impress you, I don't know what will.

Adam Ottke's picture

Adam works mostly across California on all things photography and art. He can be found at the best local coffee shops, at home scanning film in for hours, or out and about shooting his next assignment. Want to talk about gear? Want to work on a project together? Have an idea for Fstoppers? Get in touch! And, check out FilmObjektiv.org film rentals!

Log in or register to post comments
62 Comments

Wow... I was already planning on getting one. Now I feel really good about that. I am just holding off to see what they do about the metering issue.

I've been shooting with 5DMKIII for the past month. Both video and stills. You have absolutely nothing to worry about in regards to the light leak issue. At least it has not had any effect on me. I believe it has been sensationalized by the internet. However, I definitely understand not wanting to drop 3.5k on something that does have a documented issue, and if patience puts your mind at ease, then no one can argue your logic! :)

Is it just me or does the 5d mark 2 look fing amazing at iso 6400? Save the money....

How is the D800 so good at high ISO stills yet so bad at high ISO video? Uncompressed HDMI is pretty great, but what's this?

Im honestly disappointed how poorly the D800 performs. The 5DIII is better at ISO 25,600 than the d800 at 1600. 

 @twitter-60192202:disqus You have to remember that Canon still has had several years of perfecting the way their DSLR's handle video. One reason they didn't go for a big MP sensor, and only increased the MKIII by 1mp was to make it exactly 3x bigger than 1920x1080, therefore making it very simple to downscale.

And the Mark iii has 60fps....Mark IV or whatever they will call it will be more. 

Well, D800 has got a 720p 60fps too. And don't even dare to say that 5D Mark III has got a 1080p @ 60fps because that's not true.

To be honest. Canon video look a lot more smoother with High ISO but Nikon can hold details very well at High ISO. Unfortunately there's so much noise that when you edit the film, the clip will be softer than 5D Mark III straight from camera.

Damn, feels like the Nikon D800 was about to give out passed ISO 2000.

Btw that model blinks faster than my car's signal.

I'm a 5D3 owner and have been shooting with it for a good week now. I can say that the posted samples above for video is fairly accurate. This camera kicks some serious ass when in comes to low-light, high ISO performance. 

I will say though that I'm a little sad about the performance of 5D3 in low ISO from 100 thru 400 in stills. I say that in comparison to the D800 in regards to dynamic range, especially shadow details. I found that the 5D3 still has the same crushed shadow as the the 5D3. Very little has changed in that regard. With that said, it's not much of a problem if you get your exposure right in camera, without the need of bringing shadow details back in post. It's really just a small thing for me, but it would've been nice to be able to bring back shadow details in post if needed without seeing the dreaded vertical banding and pattern noise.

This is one of three reasons I went with the Mark III, So I'm glad it is as I thought in video mode.

Another reason the Mark III is better at handling noise, I think, is because the 4:2:2 that the Nikon records in, which means there is more details per 2 pixels, unlike the 4:2:0 that the Canon is still encoding in.  You can see that difference in the color bars at the bottom right, where every box is finely defined on the Nikon, where the Canon's have a smooth transition between the colors.

I'm not saying that the mark III wouldn't be better even if it had 4:2:2 video, but I just think that's one of the reasons the Nikon is noisier.

I've shot video with the 60D and the D7000, and the D7000 had less noise at high ISO in stills and video, but they are both 4:2:0.  To be fare though, the bit rate was about 22 Mb/s on the D7000, & about 45 Mb/s on the 60D, which means the 60D recorded more detail.  Now things are reverse, with the Nikon recording more image detail/data to the card than the Canon.

I still love my Mark III, but will have to upgrade when projects call for more detail, & hopefully to something that records RAW video.

How dare you post a pro-canon video on this blog! Kidding! Yes this is the one downside to a 36mp sensor. Since our cameras are only shooting 1080p video, pixel size really matters and the D800 can't keep up. The D4 is another story though :)

the ISO capabilities are excellent on the MK3 because Canon has also kept the philosophy of the MK2. They build that sensor to be a videos camera first and then added the capability of stills for it. With that direction of course you would be better video quality from the MK2 and MK3. What nikon is doing wrong in the low-light video and video in general is that their sensor is built for a stills only and then they added the video capabilities to it. In this case, nikon would obvious loose to canon because canon's 5D's sensors were made with video as its specialty, with a bonus of stills. That is why Canon revolutionized the indie and high end film production. In addition, canon still kicks nikon's ass for stills because of their glass. Nikon has to make up for that by giving their latest camera a megapixel-steriod boost. Although, I can see this pixel battle going far into the future. Canon's next 5D is going to jump up to 41mp. 

Uhm... Could you show a source for this? 

Seems like you're just making your own opinions on some comments and marketing quotes. I think at least 80% of the consumers are buying the 5D cause it's a stills camera - so it would be stupid for Canon to develop is as a videocamera. 

I do think they've put a lot of though into the video part, since the revolution with the MK II.

But clearly the girl is overall more happy in the Nikon footage on average :)

Must be a handsome photographer that doesn't need to compensate with awesome canon gear! ;-)

I'm a canon shooter, but I have to admit that you're right, Patrick!  :-)

Your right Patrick, she only smiles in the Nikon footage.  Maybe its because of all the pretty colours it produces at Hi ISO...for Nikon that is 1600 and beyond. :)

HAHAHAHAHAH, Patrick.  NICE

what bottles my mind is, the D800 would not look that bad at ISO 500 in the stills.  Actually, I don't think the D7000 looks that bad in video mode.  Also, if anything, the Mark III and Mark II are under exposed, which is when you'll see more noise.

Kinda weird.

I need to get my hands on a D800 and test it VS what is now my bro's D7000 & VS my Mark III, but with the video recorded on the card, not out of the 4:2:2 HDMI signal.

Anyone in the Atlanta area have a D800?

On preorder...

the 5d mark ii still holds its own and am impressed how good it looks at high iso and also thought it looked sharper than the iii. noise in the shadows on the nikon looked really bad. 

Something smells badly! Why nikon is so much brighter at the same ISO?

:-) IT'S SO AWESOME to see such amazing video from a dslr.
If I was into video, I'd go with the 5dMKII, to save money, have great video, and a proven performer.(and mostly because i am a cheap prick)

@tambnguyen:disqus 
 Rafal Jurczak

The reason why Nikon is so much brighter in low iso/same iso, is for the same reason why the Nikon got such a high sensor score in DxO testing: the Nikon is more biased towards the shdaow detail and Canon towards the highlight detail. . . . only in video can you see that point blank. The measured dynamic range on the DxO scores show that bias. 2 different cameras. I still say that overall the BEST camera is the mkIII LCD, ISO jpg, AF, build handling (touchy area for both camps), mkIII is a wedding photographer's dream!

btw, there is that greenish hue in the Nikon, again.

Nikon has official said that the greenish is more accurate. I'm not saying they're right, just wanted to point that out.

yeah I read about that yesterday. . . . the D800 v mkIII test in the field by pro phorotraphers shows a problem with the hues when compared to the better LCD on the mkIII. Hopefully they can get DxO to score their LCD too ;)

The 5D Mark II really kicked some ass. 

Something doesn't seem right about this test. The Nikon footage is brighter. Doesn't seem like overexposure though. It looks like it was pushed to look brighter in post which in turn causes more noise. The color bars should look the same across camera brands if the exposure settings in the camera were identical. What would cause a difference would be tonal curves which could be due to a picture profile in the camera or something done in post. Doesn't seem to be apples to apples. Despite that, the Nikon footage has more detail/resolution. Which is now making ask, is there some sort of NR added to the Canon?

Both cameras should have the same exposure (brightness) of video at any ISO (when both set to the same iso of course) !!! The difference should be only noise. I am a canon user and canon fan but I don´t like that when one side or the other is cheating in tests. Maybe Nikon is really not that good as Canon but this test doesn´t show reality. And why the hell he turned off noise reduction on Nikon? And nobody know if there was no post-proccesing after.

Thanks for iso comparison.
Right now I'm fighting  to get a clean sound with my 5dm2k.
I've tried several external mics, but it seems, that there is always too much hiss noice when recording directly to the camera.
When using the same mics on my Zoom H4, I have much less noise.
I would love to see a thorough test of the noise in Canon 5d mk2 when recording sound.IS it possible to get an acceptable noise level recording directly to the camera?

My Nikon D7000 doesn't even get that much noise in the high ISO unless processed that way, or NR is off...This test is biased to Canon, no way it's legit...

No way in hell would the ranges would be that different. Someone take it upon themselves to film a proper test that isn't biased to Canon please? Thanks. 

Anyone else fell in love?

if you watch carefully those videos are different. with dark background in canon version and light in nikon's.
you can't take it serious.

Exactly...This test is a total joke. 

Your right, the Nikon would look even worse with the dark background because this is where the noise shows most. 

Are Canon paying you guys for this? The Nikon D800 kicks the Canon's ass in low light.
http://tinyurl.com/bmkzunu

We're actually sponsored simultaneously by both Nikon and Canon. 

 That was a joke post, right????   :)

I don't shoot much video so sorry if this is a n00b question, but....why was the D800 noticeably brighter at each ISO?  Was the D800 set to 1/50 while the 5Ds at 1/60th or something like that? 

The dynamic range of the 5DM3 is impressive too.

Why is the color balance different between the 5DM2 & the 5DM3? Did they use different picture profiles.

Why is everyone surprised at this? I am Nikon guy and I fully expected the D800 to get smoked by the 5D3 when it came to video. What makes the D800 so good for stills also makes it pretty horrible for video. The D800 is having to bin huge amounts of pixels to get down to 1080p where as the 5D3 is not only binning less pixels but also has larger pixels in its cmos sensenor. In terms of video the D800 is basically a D7000 due to the pixel destiny being almost the same and having to bin pixels to get to 1080. On a similar note if you are shooting on the 1080 crop mode offered in the D4 I am pretty sure that would smoke the 5d3 and the D800. In this mode you are not binning pixels and you are able to make use of the larger pixels found in the D4 sensor. 

 A test I'd love to see...

Stanley Kubrick would've just used a f/0.7 lens instead for candle-lit scenes like these :-)

i shoot video at iso 2500 or even 5000 a lot on my mk3.. in most situations, iso 5000 looks pretty damn good blown up full-screen on my computer monitor.  best low-light camera i've ever owned.  this also gives me more flexibility when choosing my f-stop.  in certain conditions, i'm very comfortable putting my iso at 2500, and shooting with a slower zoom lens, whereas before i would have had to put on a prime lens, opened up, and shot at a lower iso.. love this camera

thats all fine a dandy but it seems to me that with the nikon you only need be at iso 600 and the lighting is perfect. with the cannon you need to boost it to 10,000 before you get the lighting to be sufficient. so if the nikon looks great at 600 where the cannon needts to be at 10,000 i dont see to big of problem

I have done my own testings last week, D800 and 5DmkIII. And a light meter, Sekonic L308-S.
Just to test the real ISO on each body. Both 50mmf/1,4 at 5,6. Speed was the only variation.
Interesting results - (DXO will have to explain how they do their own testings...)
The D800 proved to fare between a third or a half of a stop under the real ISO value (1600 displayed = 1000/1250 real ISO)
The 5Dmk3 was between ONE and 1,3 stop under... How can it be ? Is Canon cheating ? 
This video test could prove something : ISO 800 is as bright on the Nikon as ISO 1600 on the Canon. 
Exactly what my own testing (on stills though) showed. We even changed lenses several times to exclude lens variability. Same thing. 
So... My conclusion is... Whatever ! I don't give a damn about brands or numbers. Both are incredible cameras - If I were rich, I'd get the Canon for video, the Nikon for stills, the Canon's AF & body and the Nikon's ergonomics. 
OK.... Back to my Holga now.

wow.... I am super proud of my 5D mkII!!!!! dangit.... never tried it at so high ISOs before... going to now!!!

More comments