I also downloaded the original grass footage. I think I found the problem.

When I play the footage in the timeline the grass is there and it is not crushed, so you can lift it because it is there.

When you play the footage on window media player, it is completely crushed and you cannot see the grass, you just see black. It seems WMP is doing wrong interpreting the gamma when it plays the file.

My supposition to what happened to that video is when it was edited the software or the encoder interpreted the gamma in a wrong way so the original footage looks crushed when in real it was not.

So it seems it was not done intentionally, but I would never upload a video with this kind of error, I would redo until the original footage looks like the way it is before upload it. I recommend @Craftwerk to be more scientific in the tests, I mean: just upload things without this kind of editing error.

Edit post: I would like to say sorry due to be hard talking when I found the problem, I will try to be more polite in these kind of situations if I find wrong things.

@apefos "So my question is: why there is these lines in the grass original video and these lines just disappeared in the original bike videos? Both is high iso... so the difference is too big that makes me think it was shoot with different cameras."

The old GH1 did similar things...as does my cx730 at high db..this is the unreliability in shooting at high gain.

I have a question. Why would you think Hans would want to do "fake" tests?

You need to get out a bit more.

I did a test right now with the GH2. I set it to 12800 iso with a manual lens. So I changed the fstop in the lens.

when the image is underexposed the LCD shows these pulsing horizontal lines and they tend to disappear when I open the fstop and expose well. Also, even when the exposure is correct, if there is a dark object in the frame the lines appears in this object. Also the Smooth and Nostalgic shows more lines than Standard and Cinema because they lift shadows more when exposure is the same.

So to use high iso it is a must to do a correct exposure and choose a Picture Profile (Style) which avoid the lines. Neutral can give more DR to the shadows but probably it will show more lines than Standard in high iso. Neatvideo can help minimizing these lines, in my tests the noise reduction decrease the nuisance from the lines, but not erase them completely. Maybe the horizontal lines appears more in the grass footage because it was underexposed and not on the bike footage because it was correct exposed.

A good test to do with the D5200 is to try to find how much it can get underexposed in high iso before show these horizontal lines with each picture profile/style. It is simple, if someone wants to do it, just select each profile and shoot a night video in 3200/6400 iso with a manual lens changing one fstop at a time.

@Mimirsan I will get your advice! I am too nerd about gear! There is a song from a Brazilian singer which says: "I used to love as a fisherman who loves more the toils instead of the sea". I always think about this song when I am too much immersed on gear, there is a whole world out there waiting to be shoot no matter the gear you have!

@apefos, I think you should use some restrain, because Craftwerk is giving up his time for us all. There might be some errors but that does not mean that his test are not valuable. He is giving raw footage from the camera and everybody can make his own judgement. I think you are right about the dynamic range test. The original footage was crushed so the lifting of the shadows looked more than it is possible.

Looking at the jpeg chart (seems that video follows closely the jpeg values) the D5200 follows the Nikon trend like the D800. So if we extrapolate with the D800, the dynamic range could be close to 12 stop with a flatter profile. For those that could do a test please test at normal iso values also, I see so many test in low light, sometime I call it no light.

http://www.techradar.com/reviews/cameras-and-camcorders/cameras/digital-slrs-hybrids/nikon-d5200-1110231/review/5#articleContent

http://www.techradar.com/reviews/cameras-and-camcorders/cameras/digital-slrs-hybrids/nikon-d800-1074183/review/5#articleContent

@Craftwerk I just want to to express my gratitude for spending the time to do these tests. If it were not for your generosity, I don't think that many of us would have even given this camera a second look.

Regarding the horizontal lines in the shadows, I think that it could have something to do with your ISO setting. That doesn't necessarily mean just lowering the ISO to the lowest point possible. Sometimes there is an odd sweet spot. On my GH2, ISO 640 is just about as clean as ISO 160. As a result, I won't shoot at any other ISOs.

@QuickHitRecord That could be the case, I thought I saw some changes in the pattern when shooting indoors i had to raise the ISO from 200 to 400 and I believe it looked better. I'll check that one out too. But I must say that this type of noise doesn't bother me because it can easily be dealt with. For those of you who doesn't use Neatvideo, there is this free and good conversion program for video files, called Mediacoder. It has a denoiser build in that works really good.

I know the 50 isn't VR but there's still a lot of electronics in there. If you notice any difference with VR off then it adds weight to my interference theory. The only way to be sure is to try a fully manual lens.

The photo I looked at shot @ 2500ISO had absolutely no FPN (fixed pattern noise)

I'm seeing a spike in the histogram deep in the shadows looking at some jpeg stills, that's the FPN. It's being talked about on the photo forums too so hopefully Nikon will address it.

Anyone know if the clean hdmi output is uncompressed and 422.

Yeah, but does clean mean uncompress and 4.2.2

That's exactly what I'm thinking but the jury is still out on the rolling shutter and any other potential quirks.

Second clip. Raw footage straight from the camera.