Jon Holstein: I hope they decide to make a "D710", to be a mid FF model. With a nice amount of real time control, and vastly improved AF-system over the D610. If possible, a bit higher bitrate, and probably around 24MP resolution. Kinda a FF version of the D7100.
I think they can put the D610 to rest, and introduce a more entery level FF. They could skip the focus motor. Have less real-time controls. And probably lower burst-rate than the D610. Like an FF of the 5000 series. (but with two dials, instead of the single dial of the 5000 series).
And really, they should do a Df MkII. The auto-focus system really let that one down. Especially considering the low light capability of it's sensor.
The merging of 800 and 800e, plus some smaller updates, were in line of what I was expecting in a form of a D810.
Yes, but nobody yet is able to offer a FF camera for less than $1000… and it is not the cost of the AF system that is the major reason for that.
dave: Nikon is adding the features that are in the new Sony A7 series. Use a D lens and take it out of auto mode for recording video to manually adjust the aperture.
You don't need a 'D' lens, just a lens that is not 'G'. You don't need a lens with a distance chip, you need one with an aperture ring and that includes the plain 'AF' as well as the 'AF D' lenses, or even a MF lens.
There is little point in introducing another AF system between the D6x0/D7000/D5300 and the D8x0/D4 systems. Nikon doesn't need to spend money to develop a forth different DSLR PD AF system.
One can argue that nobody wants to spend $2000 and get an AF system that offers less than the 51 point coverage and thus an entry-level FF DSLR ends up in no-mans land of being too expensive for what it offers.
Higher frame rates, lower base ISO, better AF etc. are all welcome but the area were the D800 had a real deficiency is manual focus via life view due to its line-skipping sensor read out - any word on whether that has been improved?
Kevin Sutton: How can it be sharp at 600mm on FF but less sharp on APS-C?
The performance exactly in the centre will suffer from that extra 1.5x magnification. The performance in the image corners can be much better. Though usually, the fact that the DX corners sample a better performing area of the lens than the FX corners roughly evens out with the extra strain the increased magnification brings. But this can naturally vary from lens to lens, some lenses do have noticeably better corners when used on DX if the performance drop towards the corners is non-linear (ie, a sharp drop close to the FX corners). What happens between the exact centre and the DX corner will naturally be somewhere between the unavoidable performance drop in the centre and the possible improvement in the DX corners.
If a lens resolves x lp/mm, the resolution per picture height will be x * ph. And the picture height on FF is 50% larger than DX, thus, everything else equal, the resolution per picture height will be 50% larger on FF than on DX. (All things aren't equal since the sensor resolution in pixel per millimetre will generally be higher on DX cameras which helps the resolution somewhat.)
km25: I my test, the 7s was the best by about 2 F stops to the Nikon D4s, the Nikon D4s was about 1.5 F stops over the Nikon Df and because I own one, A fuji X-T1. To be honest I like the Fuji X-T1 in Jpeg, better then the Df. Below ISO the Nikons and Fuji start to take over from the 7s, not in noise, but in resolution. 16MP can be considered low by some, just think were 12MP comes in. It must take fine movies. It is too bad Sony does not offer high speed lens that are native to the camera. An F/1.8 cuts near and F stop over the cameras performance.
How do they come up with different results for different cameras if they do not do any testing? Do they multiply the number of MP with the length of camera name and then take the square root?
Carsten Pauer 2: Is there a Reason why the fresh Nikon D4S has less DR as a 6 Years old FinePix S5 Pro ?
DxO does not attempt to rank cameras based on the performance derived from one (or three) measured properties, the only rank sensors (if we include A/D converters and things like power supply). But they cannot stop people wanting to denigrate them (for their own satisfaction) from pretending that DxO's list of camera names shows the intention of DxO to pass of sensor rankings as camera rankings.
Let me ask you, how do you list a ranking of sensors without using the camera names? How do you ensure that others do not wilfully misrepresent you?
I have no idea what you mean with 'based on theory' apart from maybe that everything you don't understand is 'theory' for you.
Light composition can vary a lot, of course you have to select one or several standards and test the camera with them. The ideal way to test a camera would be with monochromatic light, and then one by one cover the whole visible spectrum. Once you know the response to each wavelength, you can predict exactly how it would react for any imaginable light source.
I get the impression that the only thing you would call 'real testing' is where You are presented with two images from two cameras and can decide which one You like better.
And you don't design lenses by building a prototype and tweaking it. You first do enormous amounts of simulations. And you largely don't use humans to optimise the lens design. You create an objective function and a good number of constraints and then let algorithms optimise the lens design. The human input is in selecting the objective function and the constraints.
That is the whole fricking point. You measure each component and each property separately such that you can compare each value separately. If you were to measure noise on jpegs, you would not be able to tell how much noise the sensor 'produces' and how effective the noise reduction is. Since you can combine cameras, lenses and raw converters in many different combinations, it is impossible to compare all combinations against each other, better evaluate each component separately. A better lens will be better on almost any camera and a better noise reduction will be better for most cameras.
forpetessake: None of those cameras are particularly useful beyond ISO 6400, and until that point A7s is behind everybody else. Going at higher ISO it looks like A7s is cooking its raw, but it's of little use for anybody but videographers.
You cannot cook the raw to extract detail that isn't there in the first place. Otherwise you'd be able to take the A7r or 5DIII raw file and cook it yourself until you get the detail the A7s is offering.
How would you express read noise (at base ISO) by anything but a single number? And no, this is exactly what DxO is doing, they measure several parameters of sensor performance and publish the results. And for those curious enough to know what a composite score for a sensor would be they publish additionally a very crude average.
Anybody who sees this differently is just intentionally interpreting things in a way that allows them to criticise DxO. Your stance is similar to criticising a plumber for only providing plumbing jobs and not those of an electrician as well which makes him useless as a general contractor.
That is the slanderous insinuation everybody criticising DxO is using. Somebody publishes a list of cameras sorted by their read noise and you bluster: How dare they publish a measured property. I guess you would also be opposed to somebody publishing a list of cameras sorted by their sensor resolution or by their sensor size.
Dynamic range is a useful property of a sensor to know, publishing it is not a 'failing', ditto for color depth. But apparently publishing any measured property of a sensor is heresy to you.
And you are only partially correct about the legal situation in Germany. The general rule is not that you are not allowed to drive faster than your tires are designed for but that you are required to use tires that are designed for the maximum speed of the car. The only exception is snow tires as for a long time there were no 250+ km/h snow tires (or they were very expensive) and thus you are allowed to mount snow tires with a lower limit but you are required to have a sticker on your dashboard with their speed limit (and you have to stick to it).
And I have driven cars which felt lousy at 150 km/h (their maximum) and some that felt ok at 200 km/h (on a steep incline) despite a maximum of 160 km/h on level road. Handling at 200 km/h requires different properties than handling at 80 km/h.
Just because one piece of information is more important than another one, the less important one is not useless. Of course lenses affect colour but so do the CFAs and noise. Your only point of reference seems to be about how something feels. You don't design a sensor or a lens by how it feels, you try to break down how something feels into measurable properties and then design to best meat those properties. And then, at the end, the fine-tuning will be done by how it feels.
I live in Switzerland but Germany is not far, when you are driving 200 to 250 km/h, whether your car has 100, 200 or 300 hp is something easily noticeable. Of course, how it handles at 200 km/h is very important as well but you won't get to 250 km/h with 100 hp (unless you are a motorcyle).
Read noise, the major determinant for DR, can be measured from the optical black pixels (or simply in complete darkness). No lens required and thus its result is independent of any lens.
You are putting up the straw man argument that because not all lenses might be able to deliver the DR a sensor can record, it would be pointless to measure the DR of a sensor. Well, some lenses just might deliver the DR the sensor can record, so it is anything but pointless to determine the sensor DR.
Your stance is also similar to saying that car engine performance (in power and torque) are irrelevant because how a car accelerates depends very much on the car itself and knowing the engine performance alone will tell us nothing. Guess what, take two cars with similar weight (and drag resistance) and you'll be able to predict the car performance to a good degree by only knowing the engine performance.
Carlos Loff: Great - I was wishing so much for this test - Great performance
Now I gave up waiting on Nikon D400 but since the Pentax K-3 is showing issues with shutter and Ricoh does not recognise them, I changed my attention to Sony A7s
Each time more convinced to go for Sony
In hydraulis, Didn't you know that when two or people observe the same problem on a given camera model (or lens or whatever), that camera model is a dud, better to be avoided?
Well, dynamic range at base ISO is dynamic range at base ISO, I don't see it as a score but as a measured property. And color depth is colour depth, a more complex property where better colour discrimination for some hues can hide worse discrimination for other hues. The high ISO score is less useful as it can be a function of several different properties (its the highest ISO value for which a number of minimum thresholds for IQ are met) and we don't know which one it is without looking at the full results but for all cameras I checked this it always was largely the inverse of QE times sensor size and since QE doesn't differ much between current sensor it was mostly about sensor size (something I don't need DxO telling me).
Some people said the Df also already had a slightly better high ISO performance than the D4 (or at least drew that conclusion from the DxO results though I'm not sure the differences aren't due to sample variation; the 18% SNR curve between the two cameras is almost matched, at the high and low end the D4 is bit ahead and in the middle, right where the 18% SNR is almost directly mapped into DxO's high ISO score, between 1600 and 6400, the Df is bit ahead).