There is no doubt that Nikon and Canon are putting a lot of effort in creating better video features in their flagship DSLR cameras. One of the most talked about new features has been Nikon's clean, uncompressed 1080p output from the camera's HDMI connector. Both the Nikon D800 and Nikon D4 DSLR cameras allow for this feature, but we've wondered just how useful are these new uncompressed files? We decided to test the new Atomos Ninja 1080p external recorder to see if these 12x larger files gave us better image quality for our own video projects.
The one big gripe a lot of film makers had with previous DSLR cameras like the Canon 5D Mark II was that the HDMI output from the camera contained compressed video and was obstructed by camera settings and menu text. Basically the HDMI output was only useful as a composition and focus aid. Nikon recognized this concern, and now you can record uncompressed 1080p footage directly to an external recorder without any display overlays (sorry, Canon users, this feature isn't available yet in the sub $6000 DSLRs). The idea is by recording externally you should have more latitude for post processing as well as sharper footage because each file contains more data and increased bit rates. The days of super pixelated footage or jittery fast action sequences are gone for good....or are they?
To test the new Nikon D800's uncompressed footage, we used one of the most affordable recorders currently on the market. The Atomos Ninja runs about $1000 and contains pretty much everything you need to get started recording video externally. We ran a few tests before we set out to film the example footage shown in the video above. We set the Nikon D800 picture setting to Neutral, turned Active-D Lighting Off (which we found did not affect in camera or Ninja files anyways, apparently it does not work with video), and made sure the footage was set to the highest 1080 quality possible.
The only settings the Ninja allows you to tweak are compression settings so we opted for the highest quality Pro Res setting to give us the largest files possible. NOTE: We did find that the Ninja shipped with a firmware that did not support the Nikon D800 so we had to update it to the 2.2 Version which fixed some banding issues that made the footage completely useless.
Below are screen captures of our result footage (click for full res). We zoomed into some of the files to 200% and 400% to see exactly what the compression and sharpness looked like on both files.
As you can see, the Ninja files have much more contrast for some reason which crushes the blacks and blows out the highlights. Even when manipulated in Adobe Premiere CS5.5, we could not recover any of this data despite having 12x larger files. I think the files do look more "polished" than the flatter D800 files but the whole point of recording uncompressed footage is so you can control the grading in post.
If you look closely at the 200% cropped images from the water fountain, you do see that the Ninja files are sharper and offer a bit more movement detail than the Nikon D800 files. However, this increase in sharpness is minimal when viewed at 1080 and would probably be less an issue if you applied a little sharpening to the D800 files. Also this increased sharpness in moving water comes at the expense of highlight detail in the water found in the middle of the fountain.
In conclusion, we were really surprised by our findings. Before our tests, we thought the Atomos Ninja would provide us files with more latitude for better post processing. What we actually found was the Nikon D800 files contained more detail. The Ninja files, while ever so slightly more sharp, had more contrast than the D800. This increase in contrast destroyed data in both the shadows and in the highlights and provided less editing latitude compared to the files directly out of the camera. The only thing we can think of that could have affected our findings was that either the Nikon D800 or the new Atomos Ninja firmwares are still not correct for complete compatibility. At the moment we cannot find any reason to pair the Ninja up with the Nikon D800 camera but perhaps it does serve a purpose for other video recording devices that have horrible native compression algorithms or short recording limits.
Here is a vimeo version which usually looks better compression wise than Youtube:
Isn't the whole point to shoot in some sort of a cine style... hence perserving as much info as possible from either the ninja or straight from the d800? I know they're currently in beta. I would love to see some tests with 3rd party image style with both the ninja and d800
I think ideally you would want to shoot with absolutely no software effects on the original footage. If you shoot with a cine style then it is embedded in the footage and you aren't able to edit outside of that....am I right?
When I speak of cine style I refer to some sort of neutral setting as mentioned by Chad above.
my point is we are recording in the most neutral setting the D800 allows so if the detail is lost at this level then no cine style will fix that
The idea of a cine style is that it is even more neutral than what comes default with the camera.
but can you install it directly on the Nikon firmware? I know Canon has had Magic Lantern for years but AFAIK you can't install other Picture Profiles on the Nikon menu system....
I would think that the recorded material on the ninja would look more like the log c of the Arri Alexa. Really flat in order to tweek it after in post.
I have to think that it is an issue with the ninja and not with all external recording devices. Maybe a better recorder like de aja ki pro would work better?
Cinestyle is a custom "picture style" design by Technicolor for Canon dSLR and doesn't need firmware modifications. It provides the smallest contrasts and smoothest gradients so after compression we are not loosing as much information as with other flat profile.
I would suspect that Ninja is using its own profile and maybe(just guessing) there is an option to install flat, custom profile.
Here is a link to forum about Nikon custom picture style (for D7000 but may work for newer).
I'd be curious to see the results if you shot with picture control settings other than 'standard'. Or even some of the custom picture control settings out there like Flaat (
http://www.dvxuser.com/V6/showthread.php?279408-Adventures-in-the-PCU) or Tassin (
http://fcp.co/final-cut-pro/news/393-shoot-flat-on-your-nikon-d7000-and-...)
I'm not sure I understand what this accomplishes. We filmed the footage in Neutral which is the flattest "cine style" out of the Nikon. We did not post process anything in Premiere so you are seeing the flattest images off the camera with the most detail. If the blacks are crushed at this level then no post processing is going to recover data that is already destroyed.
You can actually lower the contrast 3 points lower than the neutral setting out of the camera. That's and 1 notch lower on sharpness from Nikon's neutral setting is what we've been shooting with lately. Although it typically requires some correction in post it's a terrific starting point.
I think I must have misread the original article, I thought you wrote that you filmed in the Standard picture control setting not Neutral. I just re-read it. My mistake.
I far from an expert, but my understanding is that adjusting the picture control setting adjust how the data is processed in camera. The default settings in the menu (standard, neutral, vivid, monochrome, etc) all adjust saturation, contrast, sharpening and so on before the data ever hits the Atomos or CF card. So the flatter/more neutral the setting the less likely you are to end up with blown highlights and crushed blacks and you can tweak in more in post because there is supposedly more data there. These 3rd party settings I mentioned simply apply a custom curve are extremely flat.
Yeah I understand your idea but I don't think you can add those 3rd party settings INTO the Nikon D800 camera since the firmware is locked (there is no Nikon jailbreak if you will). The Neutral setting we used has the least sharpening, contrast, saturation, etc of any setting so the detail should be the greatest.
As far as outputting to the Ninja, we found that the image quality recorded by the Ninja did not change at all between Neutral, Vivid, Standard and also was not effected by Active D Lighting....so apparently those settings do not effect HDMI video at all.
Again I have no idea how you could improve the Ninja quality since no settings in the Nikon D800 are exported to the HDMI out.
Picture control settings aren't quite the same as firmware. They can be loaded into your D800/D4/D7000/...
http://www.similaar.com/foto/flaat-picture-controls/download.html
Hey great video guys! I actually thought the D4 was going to be awesome using the Hyperdeck shuttle (another external recorder that records RAW) and I ran the same style of tests and notice the EXACT same thing. Contrast seemed to be up and the file size + cost of SSDs + the hyperdeck was just not worth it at all. I will have to say that I felt the footage was slightly sharper from the clean output, but I'm talking jussssssst barely and not worth it at all. Losing that quality in the blacks/whites was just silly.
Chad - we have been shooting with only custom picture control settings on both the D4 and the D800 and the results are pretty great. Loads of dynamic range, although the D800 has the slight edge on Dynamic range, both are equally sharp.
Lee - after picking up the D4 and shooting with the hyperdeck and being so disappointed in the results I have immediately made the switch to Premiere from FCP and am quite happy. So I agree that for our needs the clean output is just kind of useless.
Thanks for the video!
Glad to see we aren't the only ones getting these results. Maybe it's a D4/D800 firmware issue but at the moment I can't see any reason to record externally.
Patrick -
I did some side-by-side tests with the D800 + Ninja. https://vimeo.com/40788982 I do a bit of video work, and wanted to see exactly where the advantages/disadvantages were with real world comparisons. If you are not going to put your footage through any kind of final color grade, then it might make less sense to use the Ninja, although I would argue that for the time savings alone (not needing to transcode footage) it's worth the $1k, at least for my workflow.
As you've seen by now, setting the HDMI output to either auto or limited gives the same dynamic range as in-camera (no crushing or clipping). But for a heavy grade, sharpening, upressing, or other effects, the Ninja has some pretty noticeable advantages. I've posted a video (with links to a ProRes file for download (go gentle on my bandwidth), as well as full-res screen grabs to illustrate (see link above).
I should mention finally that the Ninja footage post-grade looks absolutely pristine compared to the already great-looking D800 footage, especially where there's motion that begin to break down the internal codec of the Nikon. The stuff from the Ninja looks like what it is: pure, non-compressed 8-bit 4:2:2 footage that holds up mighty well to just about anything I could throw at it. Needless to say, I'm sold.
P.S. You can actually install flat picture profiles just as with the Canons, all without any firmware hacks or modifications. This is actually a feature provided by Nikon. Look up TassinFlat or Samuel Hurtado's Flaat picture profiles to get started.
Hope that clears some things up.
Good deal, I have a request. Can you guys do a test on HDMI capture cards using a DSLR, and see which ones are recognized by online streaming sites like ustream.tv, justin.tv, etc. When I had my D3s I tried using Grass Valley's HD50 (HDMI to firewire). I had to "screen broadcast" to use the footage, and even then had to deal with online streaming sites that would charge for HD input etc.
There's so much more detail in the Ninja output that you can even see the couple walking down the park in the first picture. The D800 output is so blurry that you don't even notice them.
I agree. I think that's the whole advantage of the ninja. The strong contrast on the ninja's image should be fixable via a firmware update to the ninja or d800. Based on the video review above I'd say it's a d800 problem and not the ninja.
Hehe. I had to scroll up to see for myself. Then I was all, !?!?!? Then I was all "LOL!!"
Thanks for doing that test, that was kind of fun to watch.
Great review guys. I've see these white clipping and black crushing problems before in RAW RGB video in the video production world caused by using NTSC conventions in what should be a full RGB range scenario. By that I mean clipping the 8 bit range from 0-255 to 15-235. This error is (unfortunately) really common - MS's first Xbox 360 HDMI units had this problem and they fixed it in firmware. I'm not sure who's clipping it though - the Ninja is revered by lots of pros so I can't believe it is them.
Since you guys have tons of gear can you do a test of a gray card recorded out to the Ninja and also send it out to a reference monitor via an HDMI splitter? That way you could see where the clipping was coming from. I'd love to see what you find there.
I think when the video is grading , you really can see a different ..
Please use the PIX240 from Sound Devices .....
I wonder what was used in Nikon's demo movie with the motorcycle doctor.
The Nikon demo vid was all shot with the internal D800 compression
I was just thinking about this after watching this vid. Now I guess we know why they didn't bother with an external recorder for that shoot.
great video.. i was waiting for someone to test this. interesting results-- its definitely a lot more contrasty with the external recorder, and only looks slightly sharper (probably due to the added contrast). i'm now bothered a lot less by the fact that my 5dmk3 doesn't have clean hdmi out-- it doesn't look like i'd want to use it anyway after watching this video. thanks for the test
Well I see it at two points.. This can be a very nice vacation device to skip filming on SD cards. *but who does not take a the old fashion Lappy with them to save the files on right?* :)
How does it work on Canon 7D, 5D MK II/III, 1d x? Is it the same result?
"The one big gripe a lot of film makers had with previous DSLR cameras like the Canon 5D Mark II was that the HDMI output from the camera contained compressed video and was obstructed by camera settings and menu text. Basically the HDMI output was only useful as a composition and focus aid." So far it would only even have the potential to work on the D4/D800, as the mk3 etc still don't send out uncompressed over HDMI.
Great video! Been waiting for someone to try this out. Kinda feels good, even though I don't own a D800. If I were too, I'd know that the much smaller file size can handle a larger dynamic range - so no need to feel bad about not spending thousands on extra recorders/cards/disks and hardware.
Well, the right firmware was not available yet. So this test is a bit useless. :)
You can now download the new firmware which is certified for the D800 :
http://www.eoshd.com/content/7916/nab-rolling-news-update-atomos-firmwar...
You should redo this test, I think. Thanks for your good work !!
" NOTE: We did find that the Ninja shipped with a firmware that did not support the Nikon D800 so we had to update it to the 2.2 Version which fixed some banding issues that made the footage completely useless." That is from the article above. This was done with the newest firmware that you linked.
Would you consider testing out Black Magic's hyperdeck shuttle 2? It does up to 10bit and captures as uncompressed QT or DNxHD. Id like to know if it has the same issue? It seems affordable and because the ninjas monitor doesnt do false color or focus peaking I found it too expensive. Id rather get a marshall monitor with this features and the Hyperdeck2.... if it works.
Rob, check out the new Ninja 2 just announced. With the Atom OS 3 they have introduced those features you've mentioned, in addition to a screen that is double the linear resolution. The new Ninja also has HDMI loop-through, so it's all around a tremendous value in my opinion.
There is a good explanation of this on DPreview's forums.
Basically Atomos better start cracking on a serious firmware update, and maybe Nikon as well
http://forums.dpreview.com/forums/read.asp?forum=1021&message=41205462
I guess you are talking about the crushed blacks and blown whites.
I believe this is a problem of black level and white level calibration being different on the Ninja and D800.
Please, read this thread which talks about the same issue with Panasonic GH2:
http://www.dvxuser.com/V6/archive/index.php/t-237584-p-2.html
There is a nice technical explanation of the problem:
Standard range "legal" video decoded by standard matrix means Y' 16-235 => RGB 0,0,0 - 255,255,255 . Thus "black" is displayed at RGB 0,0,0 , white at RGB 255,255,255 . The black level in the video was Y'16 and white Y'235. Virtually all equipment is calibrated to display Y' 16 as black and Y'235 as white. If your video has set Y' 0 as black and Y' 255 as white it won't decode it properly because the wrong RGB conversion is performed (most equipment can't handle or dispaly full range video correctly). So your blacks will be crushed, and whites will be blown. In order to display full range video correctly, you need to decode with a full range matrix ie.. Y' [0,255] => RGB [0,255] . Very few equimpent or support full range, and of those that do, very few acknowledge the flags in a video stream (the switch has to be done manually). Flash is the only one I'm aware of that accepts full range flag in the video stream
I believe this is exactly what happens. D800 is sending out full 8 bit to preserve dynamic range, but Ninja makes all values below 16 black and all values above 235 white.
So, unless the Ninja can be set to record full 8-bit range, or D800 can be set to output in 16-235 levels format, it will not work well together.
@freakyfritters:disqus "So, unless the Ninja can be set to record full 8-bit range, or D800 can be set to output in 16-235 levels format, it will not work well together."
Page 257 of the D800 manual: HDMI>Advanced>Limited Range
"For devices with an RGB video signal input range of 16 to 235. Choose this option if you notice loss of detail in shadows."
So unless the reviewers confirm they used the "Limited Range" setting, I think this test, review and hasty conclusions are invalid. It needs to be redone - after the reviewers study the manual thoroughly.
That is helpful, but wouldn't we really like the full range of 1-255, especially since it is only 8 bit so that more of the information is preserved?
But all of you know the time recording limitation of 29:59" of this dslr cameras , I thing that doesn´t happen recording trough HDMI. Tell me if I´m right , please
yes no limitations when using an external recorder... you can shoot as long as your external recorder has storage.
Does this work with the NIKON D7000?
check to see if the D7000 has a clean HDMI output.. clean meaning no UI showing up through the HDMI feed.
when are you guys fixing this article? you already admitted to not reading the manual and setting up the camera inproperly in the comments here
https://vimeo.com/40196608
plenty of other videos have come out proving your method flawed. you're just misleading people at this point. fix this article to admit there is no contrast problem when setting the camera per the manual's recommendation. seriously.
it's so weird to review a product when you have no idea what you're doing.
here is an interesting video that shows what a properly configured ninja can do.
http://vimeo.com/40788982
the difference is noticeable.
The only conclusion that nobody wants to admit is that the D800 internal codec is really good !
I'm getting my D800 + Hyperdeck shuttle 2 on friday so I'll be able to check if the D800 codec is really good, if the Ninja is just "not that good" or if the Shuttle is really a need...
Thank you soooooo much! Great, honest down to facts review. Just saved alot of cash there
Just sold my D4 in a trade for 3K cash and D800. Now we're rolling with 2 D800s and to be honest I think it's the best choice I've made in a while. The D4 is a beast but honestly for the price, I can't see why anyone would want it. I'd rather have a D3s for photos or a D800 for video any day.
This interview says the proper settings should fix the issue with crushed blacks and blown highlights:
http://www.dslrnewsshooter.com/2012/04/22/nab-2012-nikon-d800-and-atomos...