amanieux

as gordon laing sums it up correctly; "don't get too excited : no PDAF in 4k, no 1080p with the affordable ef-s lenses"  - canon you are still #1 at the crippling game   
 

If 8bit 1.5x crop is ok, Fuji's new XT30 is just here and its only $900. 

why do you say the sensor has to be a 3:2 ratio, most displays are 16:9, this 3:2 is coming from film era and no longer makes any sense as people are no longer printing photos on paper, new sensors should be 16:9

don't expect anything like that  from sony, they have a pro 4k camera segment to protect, the only companies that could have given us the best bang for our buck ( such as 4k raw on usb3) are companies that do not have higher priced 4k camera segment to upsell ( black magic in the begining tried but they now have a higher price range of camera to protect, samsung tried with nx serie but failed as they are getting out of the camera market). let's wait for the next new comer to flip the market upside down - the obvious thing that will change everything is an open firmware where third party developers can add features inside the camera, again samsung started that but they failed (http://www.imaging-resource.com/news/2013/05/28/redesign-your-own-camera-samsung-nx300-nx2000-source-code-released) - do you remember how expensive, crappy and limited where our digital phone services before the smartphone era with the app model, this is where we are now with the digital camera market.

If Sony stick with their usual (very careful) market segmentation, it'll have just enough limitations and lack just enough must have features such that it doesn't impact on A7 sales.  
When the full reviews come out, I'll bet there are loads of people on here and in other forums saying 'if only it had xx like the A7' or 'why didn't they give it xx, they could have done it easily'.  I'll bet we're all just a bit disappointed...

don't expect anything like that  from sony, they have a pro 4k camera segment to protect, the only companies that could have given us the best bang for our buck ( such as 4k raw on usb3) are companies that do not have higher priced 4k camera segment to upsell ( black magic in the begining tried but they now have a higher price range of camera to protect, samsung tried with nx serie but failed as they are getting out of the camera market). let's wait for the next new comer to flip the market upside down - the obvious thing that will change everything is an open firmware where third party developers can add features inside the camera, again samsung started that but they failed (http://www.imaging-resource.com/news/2013/05/28/redesign-your-own-camera-samsung-nx300-nx2000-source-code-released) - do you remember how expensive, crappy and limited where our digital phone services before the smartphone era with the app model, this is where we are now with the digital camera market.

at the moment the samsung nx500 is $497 shipped with 16-50PZ, sure it is 4k cropped instead of being downsampled from 6k but the bitrate is 40% higher so the 4k image quality should be very close for 1/2 price (the small and sharp 30mm f2 pancake lens is also down to $154) (but the nx500 has a 15min max recording in 4k, is it 30 minutes max on the a6300 ?)

i am sorry but this s35 4k combination is weird at 800+650 $ why not choose a 1300$ nx1+24$ canon adapter it is cheaper  and has better video quality, and for low light the extra stop of light you get from the speedbooster,it is equivalent to the bsi technology also giving a stop advantage for low light compared to non bsi sensors. (and for an extra 24$ you can also adapt nikon lenses  )

​and in aps-c crop it may be worse than the aps-c nx1 also BSI (we will see the comparisons i am sure)  as there will less information before compression (downsampling 5k vs 6.5k) and after compression (100Mbits/s vs 160Mbits equivalent in h264)

​I agree, I was just expressing the point in the most extreme way possible for, um, comedy effect. I think they could do a killer 4k APSC cam tomorrow, but they'll sit on it for a few months so that they don't cannibalize A7RII sales. I think we'll see the A7SII before we get the 4K APSC cam. 

OK, I don't quite get the excitement here and on other forums. Isn't NX1 at $1,500 already doing much of what a7RII is doing at $3,200? Yes, a7RII has stuff like log and higher MP stills, but then only NX1 has 120fps 1080p. I take it the jubilation is over FF vs Super 35? Doesn't quite seem worth the over 2x price difference, especially for video. I'm sure there's more, can't wait to see hands-on reviews.

This made me rethink if I should get the NX1 or not, but the price is out of my range, I could get it for 2k, but 3,2k is way too steep, so for now and the next few years, im sticking with the NX1 which is still one of the best choices for video for the price. I just would have loved the FF for stills.

​The tests Tony did were shot with well-lit subjects and most examples people showcase also feature lit or moderately lit scenes. I think where the large photosites really come into play is when you have very very very low light such as a crowd at a concert at nighttime, some wedding venues, documentary shooting at night, etc. Having used the A7s, I have yet to see anything that shoots that clean with that little amount of light. 
When scaled down , high-megapixel cameras can "combine" pixels which helps to reduce the appearance of noise but if those pixels are really noisy to begin with, they aren't going to combine into something clean.
I could be wrong since I don't have the cameras but I'd imagine the next version of the A7s will do better in true low-light compared to the higher resolution A7rII

yes nx1 is much better in low light than the lx100, even though they have same photosite area (better bsi technology and noise is reduced by scaling down the croped 23,6Mpix down to 8,8mpix, the lx100 is just a 8,8mpix crop)  : same night scene iso 800 (already too much noise for the lx100)
lx100 : https://www.youtube.com/watch?v=WVaA0840d3M
nx1 (+samyang 24mm f1.5) :  https://www.youtube.com/watch?v=4gdrFX2oFDw 
guess who wins 
i am in also to buy nx500 but i'll wait to see samples first because samsung should in all logic limit video quality on purpose to avoid stepping on the toes of the nx1
warning: 800 is the sweet spot for nx1 above that it falls appart and 3200 is terribly bad for a unknown reason 
 
 

if you have 10bit per channel to convert to 8bit per channel, noise dithering is not the best algorithm, diffusion dithering is better the typical algorithm is Floyd–Steinberg - look at the difference with the noise or "random" algorithm it is day and night :  http://en.wikipedia.org/wiki/Dither
 vs 

Andrew, the math has been explained that even though only the luma could possibly be 10-bits from an 8-bit 4K to 1080 downsample, in practice all that's going to happen is a low-pass filter and any additional luma variance will come from noise; not the same thing as capturing and directly storing 10-bits from the sensor.

From the examples shown, one advantage of the A7S over the GH4 is more dynamic range from the sensor. Folks have already tested the GH4 10-bit external output against the 8-bit Canon Cx00 and the GH4 is significantly outclassed for post latitude. 10-bits can help with tonality gradations, however if true sensor dynamic range wasn't captured, an 8-bit system with more real sensor DR will outclass it.

If this explanation doesn't make sense, you can do the tests yourself with actual footage. Thomas Worth wasn't enable to show there's real 10-bit gradations with his own custom software.

If you want to be fair and accurate, you might want to revise the GH4 10-bit luma statement until you prove it to yourself one way or another with actual footage.

Ok, so there are different explanations as to what is happening from the sensor to what is recorded.
I will use my limited understanding to try and suppose what might be happening.
 
The 5d mk3 is interesting in that its native horizontal photosite resolution is 5760 which happens to be 3 times 1920.
which means that 9 photosites are being used (or should be being used) to get the values for one output pixel.
 
The image bellow shows how this would apply to a small section of the sensor.
 
5d mk3 bayer pattern to 1920.bmp
 
As you can see there are 4 possible pixel types.
 
Thomas could be right in saying that the output is YUV.
However RGB to YUV conversion could happen regardless of how the RGB values are obtained from the sensor in the first place.
 
There are 3 rough methods I can think of for obtaining the RGB values:
 
Method 1.
The most information would be obtained by reading all photosites of a particular colour and averaging them to obtain a colour value for that output pixel.  All photosites would be used.  So for example in the top left group of 9 photosites, 5 green samples would be averaged to for the green value, and 2 blue and 2 red samples would be averaged for their respective values.
 
Method 2.
If however Andrew is correct that no calculation is done on camera,
then this would mean only 1 photosite from each group of 9 would be used for each output colour value.  Meaning you are throwing away 2 thirds of the sensor information and therefore 2 thirds of light gathering ability / noise performance.
 
Method 3.
Or if it turns out the final image still needs debayering then this means only one colour is being recorded for each group of 9 photosites.  Which means only 1 photosite out of 9 is used:      8/9ths of the information thrown away!
 
If Method 3 is the case then I would say this is quite poor.
A 4K raw frame from the 4kBMPC from would offer 4 times the resolution (even after downscaling the BMPC) and 3.1 times the light gathering ability (taking into account sensor area difference of 2.9* but not photosite size)
 
If method 2 is true then the 5D MK3 and 4K BMPC will be fairly evenly matched in terms of resolution and light gathering ability.
 
If the Method 1 true, which is probably unlikely, then the 5D would be simply beastly.
 
It will be interesting to see what the case is.