alpha604: Regarding the DR & RAW comparison at the end of the review... whille the compressed raw files warrant review, the DPReview test is also flawed by not using the base ISO values for each CMOS sensor.
The higher exposure latitude or dynamic range possible will be available while utilizing the image sensors at their native base setting. For the A7R it is somewhere between 100-160, while the A7S is 3200. If you aren't flexing the A7S at 3200 you are crippling the range it will be able to reproduce. In the interest of squeezing the most in terms of DR from these cameras proper respect is also needed for how the components are designed.
We often assume that the lowest ISO (native) is best, and it generally is. However the test images were shot at ISO100. You wouldn't seriously test cameras for comparison via extended ISO settings would you? Use the accurate native ISO of the image sensors or else you are comparing apples to oranges.
Our testing (and DxO's) lead us to the conclusion that ISO 100 is the base ISO, in the sense that it appears to utilize the minimal level of amplification and offer the maximum DR.
What leads you to the conclusion that it's ISO 3200, so that we can test our findings?
Ronan_M: "with the same shutter speed and f-number, a larger sensor will be exposed to more light than a smaller one and more light allows better image quality". Surely this is incorrect? In terms of TOTAL light, it might be true, as the total sensor size is larger in FX, but in terms of light per square inch, they both gather the same amount of light. I thought the better IQ for FX came because of the pixel density was lower on FX?
It's correct: as you then highlight.
The *total* light that goes to make up the image is greater, which is what keeps noise down (though you're right that its the same light per square inch).
This allows you to predict how much lower the (shot) noise of an image would be, based on sensor size and this prediction is generally a very close match to reality, meaning that it's playing a *much* bigger role than any difference in pixel size.
Sensor size and sensor design are far more significant than pixel size. Certainly at the sizes of sensor we tend to encounter.
MJ Fine Art Photography: Hmm... You guys seem real proud that you have proven Sony liars about the dynamic range. Just one thing to note though. In your test of the A7R vs the A7s you made one huge mistake that renders your test conclusions mute. You tested one camera at its' base ISO(A7R base ISO100) but not the A7s(base ISO 3200). Not every camera has the same base ISO or the same point at which the highest dynamic range is achieved. For a test whose entire point is determining best dynamic range the obvious first thing to figure out is the actual base ISO of the camera in question and test at that point.
@Steen Bay - absolutely right, ISO 3200 is the lowest available ISO when using the SLog2 mode (which appears to equate to the same sensor amplification as ISO 640 mode does with a standard tone curve, confusingly). But at the sensor level, 'ISO 100' appears to be base in the sense of offering maximum DR.
M Jesper: I'm European, what's a 5lbs? ^.^
Edmond Stuart - the problem presented by the '1/x"-type' naming system isn't really an imperial/metric problem, since they are simply a name for a class of sensors and no intrinsic element of the sensor actually corresponds to that length.
We try to quote the actual dimensions (or, at least, common dimensions for the *class* of sensor) wherever we can. The problem is that '7.4x5.6mm' is even harder to work into text than '1/1.7"'.
If anyone can propose a more helpful naming scheme (ideally one that's not full of decimals), we'll adopt it.
We *think* we've worked out how you'd conclude that ISO 3200 is the native ISO (though we don't believe this to be the case). Could someone provide a source for this statement, so we can confirm or disprove our suspicions?
Lofote: "Cons: ◾4K footage from APS-C region of sensor is disappointing"
Seriously? Who wrote that? APS-C in this camera is 12 / 1,5 / 1,5 = 5,3 MP.4K needs least 8MP.
It should be clear for a child that this CAN'T have the full 4K resolution in APS-C mode!
So we shouldn't point out the problem to people who aren't sitting there, working the crop dimensions out in their heads?
While Sony's website shows that the camera can output 4k from the Super35 region of the sensor, we should point out that it's not very good.
Photato: If Sony could make another large pixels camera like this for their APS-C line I'd definitely ditch Canon for good.8MP, Fast and Lite Raw files, buttery smooth low light shots, clean and sharp video !
@Photato - actually, there are down-sides to having such large pixels on such an efficient sensor.
Part of the impact on the camera's low ISO DR appears to come from having such large pixels. It means that the full well capacity is so high that the camera doesn't then have enough bit-depth to correctly describe what's going on in the shadows. (The highest Raw value has to be assigned to the saturation point of the pixel, you then work back from there).
Smaller pixels of similar efficiency wouldn't have this problem (though a better solution would be to have a higher bit-depth ADC, so that you got the high ISO performance benefits *and* the low ISO DR).
Jay Williams: Error in third paragraph. Says "may with to note" but should be "may wish to note."
Both errors now corrected. Thanks for highlighting them.
The 'Feedback' link at the bottom of the page is the most effective method of reporting errors, though.
Beat Traveller: I like the idea behind the new real world test, but calling it 'real world DR' seems a bit of a stretch to me. In the Nikon D750 review a similar test under studio conditions was called 'exposure latitude', which I think is a fairer name for what you've done here.
I'd stop short of calling it real world DR because it's not exactly clear where the extra latitude after the exposure to the right comes from. The benchmark you use for evidence of increased DR is lower noise in the corrected exposure, but because this test involves software it's biased in favour of the camera with more pixels for the software to work with. At the very least there needs to be a comparison of what both cameras look like with a 'normal' exposure, to fully show the reasoning behind exposing to the right.
Keep up the good work!
It's impossible to de-couple DR, Exposure Latitude and Noise.
We hope to publish some introduction articles to some of the issues involved - that would make it clearer.
The easiest way of illustrating the need to expose to the right (to use as much exposure as possible without significant clipping in the Raw file), is to try to find any other reference point for 'normal' or 'correct' exposure for Raw.
Joesiv: I think the dynamic range test needs to have a similarily pulled back exposure test to be anything close to conclusive.
Over expose by 3 stops, and see what they look like. Perhaps the A7S would be better than the rest, thus meeting Sony's claims, or perhaps it wouldn't.
In the past Dpreview did the wedge test, and showed raw pulling and pushing, as the dynamic range representation. This test, would be if they only extended the shadow range, somewhat half baked.
Joesiv - we based our assessment of clipping on examining the raw values in RawDigger.
As soon as you work in Raw, there's arguably no such thing as 'normal' exposure.
Exposing based on the camera's metering is based on the assumption that you wish to use the same Raw number to produce a middle grey as the one the manufacturer has chosen, for its JPEGs.
At which point, exposing to the right is the obvious way of choosing the 'correct' exposure.
In principle there's no such thing as (accurately) recoverable highlights. Once you've clipped one channel, you can't properly re-create colour in that region of the image.
Anything you can recover comes down to luck (that the colour you're trying to recover doesn't feature much of an input from the clipped channel), plus clever algorithms and a tiny differences between colour filter strengths.
If you can recover highlights with full colour accuracy, then something is very wrong with your JPEG engine (areas are being represented as clipped when there's actually full data there).
As a result, you would ideally expose such that you protect the brightest tone in the image that you're trying to capture, then pull the darker regions up (Exposing to the right).
Our current tests are based on this approach: exposing to the right, then recovering information from the shadows - this is a more accurate and meaningful way of showing the differences between cameras.
mosc: "4K footage from APS-C region of sensor is disappointing"
The sensor doesn't HAVE 4k pixels to work with in the APS-C region of the sensor. 4240 x 2832 FF gives a 1.5x crop resolution of 2826x1884 or so . In 16:9 that's 1590p not 2160p. ~1590p is barely more than half the resolution of 2160p (4.5mp vs 8.3mp)
You'd need something like a 22.1mp FF 3:2 sensor (5760x3840) to cover 3840x2160 in the APS-C region of the sensor. Like, say, yet another version of Sony's 24mp FF sensor as a successor?
The pros and cons are a list of things that potential customers might want to know.
If you look at the Sony website, you'll see a pretty diagram showing it can shoot 4k and 1080 footage using the full width of the sensor and a Super35 region.
It's relevant to tell people that both will exhibit a drop in quality (one of them significantly so, since it's 2.5k footage up-ressed), when using that mode. How much importance you give this is up to you, but it deserves a mention.
User8298355610: The camera does come with an external charger for the battery. I have two batteries and two chargers with my camera.
We weren't aware of this. I've corrected the review.
We make this point [on page 7 of the review](http://www.dpreview.com/reviews/sony-alpha-7-s/7#Super35).
ET2: No charger is listed as con, but A7s comes with two batteries external charger. It's not A7 and A7r
Here is A7s unboxing.
We don't always receive a final production box. I don't think we knew a charger was supplied. I've updated the review accordingly.
@BobHowland75 - it gets worse. We Brits have continued to use miles per gallon, decades after abandoning the gallon. And yet this still didn't prepare me for moving to the US, where they also use mpg, but with smaller gallons (about 3.8l rather than roughly 4.5l).
The pints are smaller here, too.
Earth Art: Screw the camera... Nice Turner bike!
Was this shot at Duthie Hill?
Thanks, both for your comments and your work on Duthie. It reminds me so much if the excellent trails in South Wales. It's great to have so many good natural and man-made trails on Seattle's doorstep.
Ktrphoto: I agree that there is nothing wrong with APS as a format, and that the idea that it is a step on the way to full frame is mistaken. Nevertheless I disagree with many of the statements in this review.
Buying full frame lenses for APS format is NOT wasteful, whether the user sticks with APS format, or later gets a full frame camera. In many cases (and each lens has to be evaluated in combination with each sensor that it is to be used with) it results in better overall image quality, with better sharpness and resolution at the edges and corners and less coma and chromatic aberration too.
There is also nothing wrong with the "equivalent" focal lengths of a35mm lens on APS-C or APS-H. Just because they do not work out at the "standard" focal lengths we have got used to (e.g. 24mm, 28mm, 35mm, 50mm, 85mm, etc.) does not make them any less useful.
And the 50mm, which comes in for special criticism is not in "no man's land". It is a useful short telephoto and/or portrait lens on APS.
Note that I don't say you shouldn't buy full frame lenses when you shoot APS-C. I suggest you shouldn't limit yourself to full frame lenses nor should you choose focal lengths based on a format you don't yet have.
Although I personally don't find 50mm useful on APS-C, you may well do. But you may find you struggle to find the 75mm focal length you like so much if you move to full frame. Equally, you'll be left with a lens (the 50mm) that does *something* on full frame, but not necessarily something you'd previously chosen to spend money on.
Peter 1745: @ Richard Butler
Danny Macaskill used Panasonic GH4s for his video "The Ridge".
Could you recreate this video for me with using the E-M5II instead of the GH4 so I can compare their video capabilities please?
I'm far too much of a both-wheels-on-the-ground XC rider for that. And an out-of-practise one, too.
mpgxsvcd: Olympus claims that their cameras are weather sealed. However, they only state that the cameras are “weather resistant” in the specifications. In the past that just meant that if water droplets fall on the camera when it is vertical then it will not harm the camera. Has that specification been upgraded for the E-M5 MKII?
With past Olympus cameras it meant that they might not cover it in a rain storm or even if water was splashed against it. It still may not fail if you do those things to it. However, Olympus will not warranty it if it does fail even with the “Water Resistance” labeling.
Does the Olympus E-M5 MKII offer more weather resistance than the E-M1 or E-M5 did?
We try to be careful not to use the term 'proof' in our text or videos because, as you say, there's very little clarity (and possibly little hard substance) to any brand's 'weather sealed' claims.
You can look at it from the opposite perspective though: someone further down this thread highlights that most cameras are pretty resilient, even when they're not explicitly sealed. At which point, if you see claims of 'weather sealing' as meaning: 'it'll be a bit more likely to survive outdoor use,' rather than: 'I expect this to work under a waterfall,' you'll probably not be disappointed.
I should stress that this is my personal position and is based on a combination of experience, internet hearsay and the lack of concrete warranty statements - not testing.