SimsP

Actually the message of this article is not true. One may have the impression, that based on those four scores all three cameras are more or less on the same level. But the scores only give a glimpse of the sensor's capabilities, because they are obtained at ISO 100 and that is where almost all camera sensors of the same size produce very similiar quality. As soon as you have a look at the entire test you will realize very quickly, that concerning dynamic range at higher ISO than 200, the two Nikon cams have an advantage of about 1 stop over the Samsung sensor. Keeping in mind that all three sensors share the same size this looks really impressive to me. Maybe a little bit of this is due to the fact that the Nikon sensor does have a lower resolution, which also leads to a little better lowlight behaviour, though I doubt this is the only reason. For me the clear winner is the Nikon(Sony) sensor, because in my opinion it is very unlikely to shoot exclusively using ISO less than 400, even if one always has perfect lighting conditions. Thus it is not correct to claim the sensor quality of recent Nikon cameras would still not be better than Samsung's sensor from late 2014. However things can get nasty with video, as it's not only the sensor capabilities which are of importance but also the signal processing chain and the fact that Nikon cameras still have to crop their picture (just like the unhacked NX500), whereas the NX1 for example does not. This is something the dxomark test fordn't take into account, as it only measures the sensor quality itself. More sensor area means more light and more information. So I would say it really depends on what you want to do with your camera. Video or stills.

For me the most attractive cameras are Blackmagic Micro Cinema Camera and Pocket Cinema Camera as well as GH5. This is because the MFT mount system is very versatile. You have a huge number of native lenses to attach. Due to the crop factor the entire camera system can be configured pretty light weight, but if necessary it is also possible to adapt professional lenses very comfortable via metabone's speedboosters. Especially the BMMCC is very versatile when it comes to mounting options. Although it is not as small and a bit heavier than a Gopro, it is still usable as action cam on motorbikes or cars for example and due to its expansion port it's easily possible to connect it to various platforms like drones, gimbals and so on. However you have to be a little creative when it comes to camera setup, because this is simply not a handhold run and gun camera. That's where the BMPCC fits in. And the GH5 is perfect for static landscape scenes, where it can show its strengths in resolution and detail. This is where I use my NX1 which comes pretty close to the GH5 feature-set except for 10bit output.

I think you got me a little bit wrong. I was talking about using RAW for HDR video. For most SDR projects you will never really need it, because it is not that difficult to gain 6 stops dynamic range even out of a very strongly compressed AVC file. But in case of HDR production the dynamic range easily exceeds your camera's and the more you compress the more dynamic range of your sensor you will lose. So for HDR video, RAW recording can make a lot of sense. For SDR video most of the time you only get the bigger data size and a little more detail which is not such a good deal, that's where I agree with you. So for SDR I wouldn't shoot RAW either. For HDR it has to be considered, whether the advances are worth the extra data size and if it is possible to handle it. That always depends on the constellation of the project itself. But - for HDR video - if you can afford shooting RAW, use it! 4k 8bit to 1080p 10bit is possible in theory, but it's not very effective against banding. The algorithm averages over an area of 4 pixels and as a result you will obviously get a 10 bit number, but averaging means that if all pixels have the same value - which is given in case of banding - you will get the same value again in 10 bit space. It will be a similar result to using a Gaussian filter. Thus from a technical point of view I guess dithering is still the better option.

Well part of all of this is true and part of it isn't. In the beginning everyone should answer one simple question: What do I want to get? Do I want to use the footage just the way it comes out of the camera and put it on youtube? No grading or colorcorrection. Well why buying a GH5? Save the money and get a cheap DSLM or maybe even your smartphone will do the job. If my demand for quality is a little higher and I want to grade my video a little bit, but I don't need my camera for paid jobs maybe a a6500 is a good choice. For paid jobs a GH5 is probably a better choice as it is a little more versatile. And that's it basically. For youtube or even television you really don't need better cameras. But things start to change when it comes to high quality Hollywood class video, especially concerning the recent approach of HDR for video. I've been playing with the new possibilities that you have with this huge dynamic range for almost half a year now and the one thing I have learned really quickly is: NEVER USE 8 BIT!! Although for SDR video with roughly 6.5 stops dynamic range 8 bit quantization still creates pretty decent quality, for HDR it is a huge pain and limits your grading possibilities drastically, especially when combined with an end-user codec like avc or hevc. That's why recently I've been starting to use an external recorder for my NX1 which doesn't help me with the 8 bit problem, but recording Prores helps preventing lots of blocking artifacts. Actually the 10bit output is right now the one thing I envy all the Panasonic GH4/GH5 owners for. So what kind of cameras are well suited for HDR video. Well first of course your sensor has to have a high dynamic range and this is where expensive cameras are still almost undefeated but also the cheaper cameras like A7s have a great sensor concerning dynamic range. GH5, NX1 and so on deliver quite decent dynamic range, but actually you can tell the difference to more expensive sensors in post. The next important thing is quantization and this is where all the cheaper Sony cameras as well as NX1 and Canon cameras really suck. 8 bit is definitely not enough for HDR and you will get a whole lot of banding if you're not very careful. In post I very often use a grain overlay on my NX1 footage to dither the banding a little bit and it helps, but it's far from perfect. So when considering HDR 10bit recording is absolutly minimum. Better use 12bit or 14bit if possible. The final really important thing is the way you store the data. Here the Blackmagic cameras clearly have a huge advantage because of their RAW capabilities and my BMMCC has served me very well during the last couple of months. Using RAW you can make the most out of the sensors dynamic range. Prores or DnxHR also do a quite good job but you will most of the time lose at least half a stop and this is something you definitely do not want for producing HDR content. So whenever you can afford the datasize SHOOT RAW and NEVER EVER use AVC or HEVC they will betray you. Actually this leaves not many cheap cameras in the pool. The only cheaper cams for RAW are the Blackmagic ones and the next would be FS7 price tag and beyond ARRI/RED. If I had to choose a camera for a professional grade HDR production never mind whether for youtube, TV or cinema, the cheapest A cam I would go with would be a Blackmagic 4.6K and maybe a BMPCC/BMMC or perhaps GH5 as B or action cam. Every other camera below has too many limitations in one way or another and obtaining good HDR footage will be very frustrating. But the price is worth the effort. HDR with its huge dynamic range allows for a whole lot more creativity, when it comes to playing with lights, shadows and colours. In my opinion HDR is the one big thing during the next years, that will allow us to be even more creative in pictorial design than ever before and hopfully it won't follow 3D down the stream. So, as mentioned before, the question always is: What do you want to get? For common youtube quality you don't need a camera for 1k+ bucks. But if video quality is substantial to you, especially when you're considering HDR video, the range of small affordable cameras becomes smaller and smaller. Thus if you don't want to buy a new camera every other year - choose wisely. And maybe Blackmagic has something new for us in their pipeline for the not so distant future. Kudos to them for democratizing video world like never before!

I would also suggest A to be video and B the RAW picture. Why? Because of the noise in the darker areas of both pictures. B has more uniform distributed noise just like I would expect it from a RAW picture. A has in my opinion more inconsistent noise due to compression of the video file and in some places I think I can even see compression artifacts.

Thanks a lot for the review Andrew. It really helped me with my choice for my next cam. The only question I have, are you really sure the internal recording is 4:2:0 ? I asked Samsung Germany and they told me that recording is 8 bit 4:2:2 both internal and via hdmi. I don't mind because I use an external recorder anyway if I want to get good quality, but that would mean Samsung support is wrong (once again). Concerning h265 I guess we will see wide software support in 2015 when the bda announces the specifications for 4k bluray.