DxOmark (Click here) posted the full Sony A7s sensor test and crazy High-ISO fans will be happy to hear that Sony A7s sensor is the best low light sensor tested at DxO! It easily beats the second bets camera, the Nikon DF (which costs $500 more!). The A7s also prefers more Dynamic Range at High ISO than the Nikon.
DxO writes:
Sony’s A7 and A7R models have not unexpectedly generated a lot of interest with enthusiasts, and the new addition to the range looks equally intriguing, if perhaps for different reasons.
While it’s true the short register and low pixel density makes this model potentially more suitable with rangefinder type (non-retrofocus) wide-angle lenses, there’s little doubt it will appeal to those who value extreme low light performance over high pixel density. And then there’s the additional attraction of video, with 4K as an option (albeit with a third-party recorder). Even the price doesn’t look too bad against the Nikon Df, which is not only more expensive but can’t quite match the Sony’s low light capabilities and shuns video capture completely.
Sony A7s will ship soon:
Sony A7s preorder in USA/CA at Amazon, Adorama, BHphoto, Sony USA, Sony Canada.
Sony A7s EU pages: Sony DE, UK, ES, FR, IT, NL, BE, AT, CH, SE, NO, FI, PL. And Wex UK.
0 Responses
This is great but where are the 15 stops of DR we were promised?
I’ve no idea where the promise of 15 stops of DR came from. There is extra DR, and it’s at the higher ISO end, which will make it more usable in dimly lit environments – which is what this camera is about. For the HDR enthusiasts, back to the old techniques. For the vast majority of photographers requirements, I suspect there is plenty of DR at base ISO in all the A7 cameras.
Interestingly, the DR gains at high ISO over the A7 (at equal output size) are rather higher than the 1 stop implied by the DXO “sports” marking. It looks more like 2 stops, but I think DXO set their criteria for low ISO at a point before the measurements diverge more strongly.
The 15,3 stops DR comes partly from the firmware update that Sony released a few weeks back. This is part of the firmware description:
– Dynamic Range extended to 15.3 stops as sensor RAW output
It is also in a lot of their promotional material. See Amazon.com for instance.
It has nothing to do with S-Log2. If the sensor can’t output 15 stops DR in raw then there is no way S-Log2 is gone magically make it appear.
I know nothing about the way DxOMark tests this stuff, but I assume they would be using the latest firmware version. How Sony, a premium sensor maker, can be shown to be 2EV off in max DR from their own measurements is completely incomprehensible to me.
How could DR increase 2stops merely by upgrading firmware?
I think it is extended ISO such as ISO 50 or 25 but not original ISO
It’s because the original Sony raw format, with 14 bits linear equivalent, can support no more than 14 stops of dynamic range. A different raw format with a different transfer function and/or more precision is required to support greater dynamic range.
Both the A7 and A7R already exceeded 14 stops (at 8 MP). The way DXO measures DR, the theoretical maximum, even with 12 bits, is already 13.79 EV at their quoted 8MP. At 16.39 MP, that’s already 14.31 EV, etc .
You forced me to go look up how DxO does its measurements. 🙂
Dynamic range is the ratio of the saturation point to the lowest luminance level with an SNR greater than 1 dB. http://www.dxomark.com/About/In-depth-measurements/Measurements/Noise
All ratios are normalized to an effective 8 megapixels by multiplying them by the square root of the ratio of the sensor’s resolution to 8 megapixels. http://www.dxomark.com/About/In-depth-measurements/Measurements/Noise
DxO is exactly right to realize that the noise of each pixel will generally go up as the pixel count goes up. It’s fair to normalize to a certain resolution. But the way they do it is a mathematical abstraction. If their description of the calculations is accurate, they’re not actually downsampling the image to 8 megapixels and then finding the 1-dB SNR luminance level. They’re finding the 1-dB SNR luminance level of the native pixel size, and then un-handicapping it by a factor which would be equal to the reduction in noise by averaging if the image was flat (uniform color) and the noise was Gaussian and not limited by the raw encoding scale. But that doesn’t match reality. First, not all noise is Gaussian. And second, this approach could yield an 8 megapixel normalized 1-dB SNR luminance which is lower than the camera is actually capable of! (i.e., luminance levels lower than the camera can distinguish from total darkness) The greater the sensor’s resolution, the greater the un-handicapping factor, and potentially the greater the difference between the camera’s actual bottom end of its recorded dynamic range and what DxO reports. There’s no theoretical maximum dynamic range with DxO’s method: higher resolution will always help the reported DR. Well, this is disappointing. DxO just lost a lot of credibility in my book. No wonder we are seeing 14-bit linear raw cameras with DxO-reported dynamic ranges greater than 14 Ev.
Now, how Sony is claiming 15.3 stops, I have no idea. They haven’t said how they measure. All I know is the Sony ARW2 format uses 11-bit samples and has a tone curve embedded in it, and in theory the right tone curve could easily express much more than 14 or 15 stops of dynamic range. I just don’t know if ARW2 is flexible enough to let you specify such a tone curve. But I guess Sony controls all of this and there’s nothing stopping them from creating a new version of ARW2 with more precision in the tone curve specification. http://www.rawdigger.com/howtouse/sony-craw-arw2-posterization-detection
There is a maximum for their method at their reported 8 MP, using the smallest value (noise floor) possible, larger than 0. They even quoted the number I gave in their forums.
http://presscentre.sony.eu//content/detail.aspx?ReleaseID=9810&NewsAreaID=2
I think the 15 stop of DR come from the S-log 2 which apply only for video. Since DxO only test the cam for still photo you have to wait for test from videographer.
the 15 sptops are with the S-Log gama curve, not in normal gama, so must be tested with S-Log
how does it easily beat the DF but has a lower overall score?
I think the headline is emphasising the low light/high ISO performance. That said, the difference is nothing like as large as implied – perhaps 1/4 stop where DXOmark set their quality criteria for low light performance.
However, what the scores don’t show, and the detailed graphs do, is beyond that point, the gap between the A7S and DF (in dynamic range) grows considerably. That is, the performance gap at 6,400 ISO and above is considerably larger.
Clickbait.
The same way the Miami Heat Easily won the Championship over the San Antonio spurs. LOL
I prefer the d800e or the upcoming d810 (improved low light performance like d4s – we’ll have to wait and see when it gets tested ….) due to overall performance resolution advantages, better lens line up, also 2.8 lesn if it is needed…video a7s is clearlyl superior but without 4k internally it is a now brainer should have wait 6 months later in order to fix the overheating problems or design a slightly bigger body…..in order to get it done internally…..huge bummer….
Will the upcoming 810 offer internal 4k recording?
for sure no! 810 will be mid update slightly imprvements…..not substantially sadly…..
Only one which could probably have fullframe, IBIS, wifi/agps built in,af touchscreen and 4kinternal recording is the Sony A9 and or A99II !!!
But maybe the admin knows better….;)
yeah admin what do you know about the both flagshisp of the FE Mount Mirrorless and of the A Mount SLT ??
Sony A9 and A99II?
When they gonna come?? What are the rumored specs? Please a brief sum up and update for the followers……we need feed….;)
The a9 could have an integrated grip with an extra battery socket, since a flagship is almost sure to have dual bionz x and the current battery doesn’t even have enough juice handling one. But it’s never gonna be perfect since Sony’s kinda made it clear FE bodies will not have IBIS…honestly, with an integrated grip’s room, FF IBIS would be totally practical.
I’d like to see a more rugged exterior- and everything you stated.
Of course. Weather sealing always helps.
if the sony a9 and or the sony a99 II will have 36 mpx sensor of the sony a7r/d810 and 4k internally…..with decent low light performance as the d810…..wifi/gps built in, IBIS and af toucscreen like the omd em1…..
it is the ONE to get….will shaking the whole industry…..all desired features and advantages of the different formats and systems in one package….
A DREAM would come true!! 😉
Lets hope and pray! Time will tell.. we’ll see….
me too anyday ;)! big fx lens make more sense on bigger bodies…..balancing issues
sony a 7 series /fe moutn only makes sense if you onyl use or need the primes…..otherwise you need the batteriy grip makes it as big and as weighty as the fx dslr…..in order to balancing well the f4 zooms or the adpted 2.8 zooms and for the latter you need then an additional adapter with weigth as well…..non sense…..
Yep, that 160 gram adapter really brings some huge balancing issues.
Sigh… These people are at it again… Maybe there’s something wrong with their left hand so they’re obsessed with always lifting half the weight with their right.
You prefer the D800e over the a7s because of ‘resolution advantages’? Wow!
Hahaha!
3700 to 3200 is not “easily”.
I would say that, with this result, the Nikon Df may hold a better value IQ wise since it can pack a bit more MP in its sensor.
thought the same my friend….! 😉 And the df 2 is already in the pipeline due to the fact that the df is doing quite well in sales ! Surprisingly dont like it too much deficiencies…
sorry for typos…too fast too lazy,
The scores are normalised to a common output size, so if this was “per pixel”, the gap is a rather bigger in terms of DR (and what the DXO mark graphs show, if not the score, is the gap is notably wider when you get beyond ISO 6400. That said, I rather think the A7S is aimed at those who have both video and still photography needs in a compact package.
“I would say that, with this result,…”
Yes, A7s is for video mainly. That much is clear from Sony’s marketing. A lot of people can still consider this for photography, so any review or comparison from stills still applies to them.
+1
And more native af lenses with faster aperture
Up to 2 stops difference in shadow noise (below 10% gray) as you move up the ISO scale, see full SNR curves in the full reviews, but the DR curves already give a hint.
well with that light leak, will be fun with 43124124 iso XD
Not quite the DR monster everyone said it would be. The 36MP sensor still looks like the best tradeoff. Similar noise performance after a rescale down to 12MP but higher DR and much higher resolution at typical ISO speeds. The 12MP sensor would be better served in a camera body that can shoot sports at high frame rates. It seems the A7s was only made to debut a new sensor without properly weighing practicality.
totally agree with your conclusions…;)
Looks like even lower DR at ISO 50-200 than the A7. So it’s better at low light but worse at normal/bright light
Sony decided to implement a different DR curve for the A7S. They sacrificed some low ISO DR so that the high ISO DR could be increased. This likely also explains the slightly better high ISO scores.
it’s quite practical: Sony Imaging gets some money from pixelphobes with very little investment in R&D.
DR is clear disappointment. It seems that it is very hard to combine low read noise and big pixels. Such a pity.
“easily beats” LOL!
No amount of such spin would justify its price and make me buy one, especially given the fact the A7s lacks many of the features in a real camera like the Df.
…such as?
Check the specs (nikon site) or various review outfits. fps for one, for example, battery life another…….
Oh I’m afraid I’m gonna have to agree with you on these two, failure to change to a more adequate battery when moving from E to FE is one of Sony’s most critical design mistakes IMO, and only one additional battery in the optional grip? With so few dials on it considering the size there’s clearly room for two. And the slow burst…room for upgrade I guess, I’m waiting for a8/9 to jump on the wagon anyways.
Though I don’t think df’s worth the price either.
If it’s too much trouble for you to carry extra batteries than get a proper dslr with and ovf that doesn’t suck the battery life. I prefer the lighter weight of a mirrorless with an evf. The extra batteries I have to carry around isn’t even noticeable compared to bigger, bulkier dslr with heavier lens. Everything is give and take.
But the low resolution really hurts it
I had 6 batteries with my A7r a few days ago for a few hours of shooting. Barely made it with the last one at ~20% with all the others completely exhausted. I had another (M cam) with me as well. That battery is still at 70% though it has not been changed in a few weeks. I understand the power requirements for the EVF/LV. For a 12mp cam one would expect better. What is with the fps, even though the output is cRAW and from only a 12mp cam?
I’ve got 14 batteries total. And even with the grip fitted you’re effectively running a power-hungry EVF off the equivalent of an LP-E6 Canon battery. I don’t chimp and can get through one battery in two hours easily. The only way is ON/OFF with the power switch between set-ups.
A hard button on or near the hump (per Fuji) for LCD off would help, then when camera is away from the eye the EVF would sleep too saving power.
Note: Since the firmware update the LCD appears to be completely OFF when EVF only is selected in the menu, this item should assignable to the VF menu but it isn’t!
I agree with you, the little TV/EVF is a power-hungry sucker — but its a beautiful one!!!! (:-)
It’s really not, at least in this case.
Smaller batteries mean you have to charge more batteries for the same amount of power, and you have to charge them one by one, that’s extra work that could be avoided.
Changing batteries more often means pausing more often, what if the tiny battery forces you to pause at a critical moment? Not to mention more batteries mean a higher possibility of losing one, and maybe forgetting to bring one with you.
How many have been complaining about the small grip of the a7 series? A bigger battery gives you a bigger grip.
You’re misleading the conversation. A bigger battery gives you more power, EVF or not. To save power I could still use the screen, but still a battery meant for half frame only gets you so far on FF.
Good points, planning your shoots is critical now. I think its pretty cool though, having a camera like the A7R with MF resolution, better than pro analogue audio and near broadcast quality video that’ll run on a tiny battery for a couple of hours!
I can remember video cameras that were umbilically attached to battery belts that’d have Homeland Insecurity Agents diving under their desks…….(:-)
Higher burst will come with the A7s mkII. They have to make a some mileage out of the “new” sensor. 😉
Oh god no, since the current design is already ugly enough, a different design should mean a better design, if an mk2 comes out it’ll look the same…
Seven s seems junk. Go for the r
Not if you are a multi-media shooter.
Well, it does beat it when it comes to low light according to the test, so not much of a spin.
“Real camera” is a subjective term. If your’re one of those old curmudgeons who can’t accept change then yeah, this is more of a real camera. If you can handle adapting to the times, then this Sony is much better. The A7s does amazing video, the DF can’t even do that and they charge you more to take out the video feature. I’m sure the DF is a nice camera but I prefer the specs of the A7s much better.
Now I agree with you
Df have the same scale of video shooting as D700, and that is what A7s cannot reach, REAL CAMERA, LOL!!
About 2 stops difference in shadow noise at a measured ISO 33k for the Df (A7S scores about the same at ISO 150k), could be called ” easily” if one is interested in just low light shadow noise.
LOL you are stretching the old fanboy routine here , it may well be better than the DF but honestly the high ISO files above 12800 ISO are atrocious quality just a bit less atrocious
This is what the shadows look like at 104000 ISO
Cute attempt Bob, but I could care less about your OOC jpeg crop. This whole topic was about RAW, get with the times.
God , how much are Sony paying you lol , do you not get embarrassed making so many fanboy posts.You are still in denial about the weak high ISO of the A99 compared to other cameras using the exact same sensor ,despite every review site on the net pointing it out
Imaging resource have RAW files for the A7S here is the 25600 ISO RAW compared to the DF RAW using IDC for the A7S and ACR for the DF .At the same output size the difference is at best miniscule. Above this ISO the image quality is frankly unacceptable on all cameras
I downloaded the IR raw samples ,personally the advantage to the Sony under 51200 are pretty insignificant. For me it is all irrelevant as I never shoot above 800 ISO so I am more than delighted with my A7r. I guess if you regularly shoot video it 150000iso the A7s is the tool for you .
I’m not in denial, I actually shot the A99 side by side with the D600 using the same T-stop and shutterspeed in low light and know the difference from experience (nothing to lose sleep over) and posted the results. You go by some odd OOC jpeg crops from the A7S (which firmware?) and studio scenes with often differing amounts of light and to top it, completely different RAW converters (IDC isn’t even close to being in the same ballpark when it comes to handling noise and detail, compared to Adobe converters). And then I’m in denial and a fanboy for pointing out your obvious nonsense theories.
Good one, keep the nonsense coming, if only for our amusement.
In the meantime, this is how the A7S really compares to the Df, both converted with Adobe converters, normalized to the same output, no luminance NR:
http://horshack.smugmug.com/photos/i-qXh4grb/0/O/i-qXh4grb.jpg
And the Df actually received 1/6th EV more light too, suffers more from (horizontal) banding, has a lot more shadow noise and a loss of DR (blacks are no longer black) and colour shifts.
The “Larger photosites produce higher signal to noise ratio and brighter/cleaner pixels” people:
“Yeah,..no real news here. They are confirming something we already knew”
“We have been asking Sony to make this sensor years ago”
“Sony FINALLY made a sensor for us big photosite, low light shooting fans”
“This sensor generates some serious voltage! (12 million high voltage pixels!) Sony should wire the sensor to the battery. Point the camera into the sun and charge the battery at the same time.”
The “Sensor size matters, not photosite size” people:
“What? How could this be true?…it must be a LIE!”
“I rather have a 36mp sensor that looks to only have a bit higher resolution than 12mp any day”
“I don’t understand. I’m sure my 36mp will look just a good scaled down to 12mp”
“My mommy tells me that 36mp is higher number than 12mp…so it MUST be better right?”
“This ruins everything I used to believed about high pixel density…there is no God”
The reaction to the A7s is so polarized between the people that are stunned and surprised and the people that say “We told you so”.
Good job SONY!! You easily got my $2500 on this one. the A7s is the most interesting camera you have made in years. Cant wait to start shooting with it.
?? Look at the measurements at DXOMARK. Scaled down to the same size the difference between D800 and A7s is about 0.4 EV at ISO6400, similar at 12800 for S/N.
Dynamic range up to 1600 in favour for the A7r/D800 thereafter for the A7s.
So one has the choice of what is more important for one’s shooting. I’m with the higher MP.
Me too now.what a big disappointment this low resolution chap is , it must have been much better than the seven r to justify the high premium over it.
The actual resolving power of the 36mp A7r is really only marginally better than the A7s and A7s. 24mp file sizes may be twice the size of 12mp but don’t resolve even close to double the lines of resolution on a chart.
The higher and higher you go in megapixels, the law of diminishing returns gets stronger and stronger.
It’s certainly not marginally better, but a lot better (1.73 x in each direction). If you can live with ISO 12800 for stills, a downsized A7r picture delivers very good image quality (S/N almost the same, DR 1.2EV better for A7s, 0.7EV at 6400). Things will be much better with newer generation sensors and less photons lost at sensel boundaries.
Shoot an actual resolution test chart and you will see that the two cameras are not as far apart as you’d expect.
I actually often shoot my Xt1 alongside the D610. I see a difference there, already. So I guess it should be more pronounced vor 12 vs 36 MP.
I don’t know what you are so happy about. Yes, larger pixels produce better SNR, but so does the larger sensor. Its both! Otherwise RX100s would easily outperform last gen APSc sensors.
Within the same sensor size, where you gain in ISO, you lose in DR and Color Depth. Take A7s, Nikon DF, A7, D800 and A7r in comparison. Its clear that for the same sensor size, lower MP count produces cleaner files, while higher MP count produces better DR and Color Depth.
As a side note, looking at the 2nd part of your post, you need to grow up….
Why does higher pixel density cause better DR and color depth?
I can only make a guess based on multiple tested cameras. My guess would be is that due to a higher pixel count, in a bayer array sensor (the key point), the firmware has more data to interpolate into an image. So essentially more pixels are available to translate the same color.
Again, this is a guess based on multiple tested camera numbers from DXO. I’m sure Stacked/Foveon sensors perform in a completely different manner. There a larger pixel might actually be better in DR and Color Depth. But who knows. Unfortunately we don’t have too many recent foveon sensors available, with latest one being from sigma and its already quite old (2007 from SD14?).
We can only wait and see if Barry’s hope for stacked sensors will materialize in 2015-2016. Only then can we do a proper comparison of latest gen sensors.
Even if they manage to reliably and cheaply manufacture APS-C sensors with the same pixels used in RX100M3, (65 megapixel on aps-c) that’s be quite good
Because of oversampling, that every electronics engineer should know.
Read any electronics textbook about Analog-to-digital converters.
Ok ok…I’ll admit that much of my excitement also comes from the A7s video capabilities. Full sensor readout with no line skipping? 4K 4:2:2 HDMI out? Full knee and black level control? S-Log and other gamma and color controls?
This thing is a video monster too. (That’s where I make most of my money anyway)
This might become the astrophotographers camera of choice?
Sony’s lossy RAW produces artifacts around high contrast transitions (white stars on black background) and antagonistic with necessary dark frame subtraction, so astrophotographers don’t like Sony.
I’ve seen amazing astrophotography with existing Sony cameras such as the A99 and A7R already, despite the exaggerated cRAW claims.
A7s is a bit better than A7r at high ISO, A7r is a bit better than A7s at low ISO. Not a big deal either way.
Yeah honestly disappointed , I thought it would be much better than this , other wise, who wants to buy a low mp camera of this high price?
videographers?
Videographers would have bought this camera if it had internal 4K…
As it is A7s may be a failure from a sales perpective. I do not say it is not a good camera, but is has no specific target customers.
Exactly. What was Sony thinking?!
Videographers????
perhaps an amateur “videographer”.No pro is going to shoot video with a still camera. They’re a joke for most video shooting.
Tons of professionals use DSLRs for a variety of work. Try looking up Philip Bloom. http://philipbloom.net/
Ye kip dreaming, your comment reflects trolling and lack of knowledge. There are so many examples, for instance in transformers they used Canon 5D mark 2 for fast actions mounted on the cars (it’s image quality is a joke now) but then it was good enough and is was small.
Most serious videographers will prefer to use the Gh4 with extertnal recorder too for 4K, since it’s specs for onboard 4K (100Mbps IPB…, 4 times the pixels of FHD) are rather limiting for some good grading.
In other words, many are used to using external recorders for uncompressed and best output in general anyway.
Most videographers still shoot at Full HD, and the Full HD on the a7s looks much better then the Full HD on the g4 even the 4k on the a7s with lower bitrates look much better then the g4 wich looks like the video from a typical camcorder but with much more sharpness vs the filmic look of the a7s that most people prefer-
This comment is a joke.
Sony A7s is a failure and the sales will reflect it! Not everybody cares about ultra shallow DOF, in cinematography to get the shallow DOF they use tele lenses, I never heard of people shoting at f/1,4 on a FF sensor camera (not on big productions films anyway). Most of the cameras are super 35 like (Aps-c size).
Every time somebody is commenting about the lack of 4K in A7s trolls jump in defending the crippled product. Yes it is a crippled camera, no trolling is going to make it better. Sony will use this sensor which is phenomenal bay the way in better and more expensive cameras in the near future. I’m sure of it.
Yep, if you’re primarily a stills shooter, I still think the regular A7 is the best compromise of features (I’d pick it over the A7R because of EFCS.) For the price, the A7 is particularly appealing.
sony will always be lowest of the low, or as we say: последната дупка на кавала. 😉 period.
case closed.
You mean lowest of the low in terms of super clean low ISO files.
Lowest of the low, as in low noise per pixel, low failure rates?
Lowest in audio pre-amp noise?
Interesting on the DR. Sony states 15.3 stops but here it states around 13. What gives?
I look forward to the RX2. 🙂
I’m specifically not buying the A7R now because of the rumored RX2. I’m that interested in it. One crazy good lens, an amazing new sensor. If it has better AF than RX1- I’ll be the first in line. Freakin can’t wait.
Dynamic range is defined as the difference between the highlight clipping point and the level at which details is lost to noise. Depending on how you define the “level at which detail is lost to noise” you will get different answers for the DR.
DXO arbitrarily define that point as an SNR of 1. It seems Sony use a different definition (somewhere around SNR 0.25)
The announced that as a firmware upgrade about 2 weeks ago and is not available yet.
http://presscentre.sony.eu//content/detail.aspx?ReleaseID=9810&NewsAreaID=2
When can we stop looking at DxO numbers? Their hokey, black-box tests do little to tell u how cameras will actually perform.
Yes and No. We clearly see what downsizing gives us in terms of dynamic range and S/N for different ISO. So it helps in making a choice. Of course we should supplement this with real-world tests and incorporate other factors like lenses … For this I look at your site, for example 😉
What a ridiculous statement. DXOmark scores correspond very well with technical performance. If people care to look beyond the simplistic scores and look at the actual measurements, then it becomes clear in what circumstances a camera will excel and when it is poor. A look at any of the sites which performs properly controllled
Of course, DXO doesn’t tell you everything about a camera’s performance. It won’t tell you about AF performance, controls and ergonomics. It won’t tell you about anything about the output from RAW conversion (either in-camera or external). However, as a set of objective measures on what a sensor is capable of capturing, they are extremely useful. Producing camera sensors is a branch of engineering, not some magical alchemy.
What’s clearly visible from the DXO numbers is that the A7S excels once you get beyond about 64000 ISO, and if you want to photograph in the dimmest light conditions, this is the best there is. However, at lower ISO levels its performance is comparable to most current generation FF cameras using Sony sensors (Canon sensors are a different issue; OK at medium/highish ISOs, but losing out at base ISO).
If you don’t find this useful, then fine. But some of us do, especially as it provides an element of objectivity.
their data is far more “facty” than objective. Until they open the lid on the black box (which they will not do) – one must be incredibly cautious about drawing any conclusions from their tests. How many people look at these reports, and make “better/worse” inferences. Even the measurements cannot be linked to least perceptible differences. It’s numerical, standardized – but still obscured to some degree.
And let’s be honest – the only engineers who would use DxO data for anything but entertainment are the armchair variety.
I’m not sure why you think the DxOmark testing protocol isn’t well defined. There’s quite a lot of information in reasonable depth on their website. They’ll even answer questions. The only thing that’s a bit mysterious is the point scoring system, and I think people would be well advised not to treat those (particularly the overall score) at all seriously. It’s far better to understand what the metrics mean.
There is an aspect where people do obsess over what are really very minor differences which are unlikely to show in reality. For the most part, all modern sensors are pretty good and the technology is now fairly mature with incremental improvements. The low light performance of the A7S is more a result of choosing characteristics optimised for that purpose than any major technological breakthrough.
I think there are more differences in things like image processing, intelligent AF than on IQ.
+1
Who cares – just get any camera and go take pictures. Like an iPhone:
http://www.dpreview.com/news/2014/06/15/2014-iphone-photography-awards-winners-announced?utm_campaign=internal-link&utm_source=news-list&utm_medium=text&ref=title_0_8
So it has worse color depths, but 0,1 stop advantage in ISO performance. Yeah, definately beats the Nikon “easily”
Looking only at peak numbers and creating an opinion solely on that is short sighted. While the Sports score may only be around 1/6 stop advantage, look at how much more dynamic range the camera has at the higher ISOs than the Nikon has.
The Sports score is based only on the lowest common denominator of 3 values from 3 different measurements (SnR, Color Depth, and Dynamic Range) at a given ISO using their print view. SnR can’t fall below 30dB, Color Depth has to be at least 18.0 bits, and Dynamic Range needs to be 9.0EV or higher. If 2 out of the 3 values meet or exceed those requirements but the third value does not (even by a fraction), then the camera’s Sports score ISO is then rated on the lowest of the 3 values. Thus, it doesn’t paint a clear picture of its true capabilities.
At a measured ISO 33k, the DR of the Df is similar to the shadow noise of the A7S at a measured ISO 150k (just 0.2 EV difference). The full SNR curves (not just those for middle gray) paint a similar picture with up to about 2 stops difference in shadow noise as you move up in the ISO range. And shadow noise is usually the limiting factor in low light (lots of deep shadows).
nevertheless…af system sucks heavily or is jsut insanely bad…..sony a7s and sony a7r……..d800e much better….
they just should put in the af system of the a7 in the a7r and s….and everything would be fine or at least much better…..unbelieveable and absolutley not understandable…i still think the ff dslr companions il have the edge in af…..especially CAF….but that is not so important for the majority of users…
so we have to wait for the successor of sony a7, a7r,s or higher class models a8 and or a9 to have 36 mpix and better af system….
Actually, some people are reporting better AF performance from the a7R (cdaf only) compared to the a7 (hybrid af). What we need is Sony put the a6000 AF system into practice with other models.
I’d rather like to see DFD AF (like Panasonic does) introduced in all models (including older ones) via firmware update
For a camera hyped for high ISO the SNR results are nothing special even compared to the D3s . The bottom line is that the only gains { 1 stop +} are for DR at ultra high ISO north of 12800 ISO
> For a camera hyped for high ISO the SNR results are nothing special even compared to the D3s
No wonder 🙂
This is a fools errand. Both are good.
Sony better build better FE-mount glass before anything.
I use both Sony and Nikon, and the new Nikkor G series f1.8 primes are a match for the Sony Zeiss FE primes in real world use, and cheaper too.
I’m not complaining, just pointing out Nikon has some outstanding new lenses.
totally agree…..but only the primes are excellent and we still only have 2!!!!!
and the only fact which makes sense to use the sony a7/investing in fe mount…..a7r not for me at least (shutter problems and horrible af, handling like mf body physical stabilization is better if you have a bigger body so d800e woudl be always my choice, a7r would make sens or i would choose it if it has IBIS of the omd em1!!!)
the standard zoom 24-70 f4 zeiss which would be or will be the “bread and butter lens” for a lot of pros and enthusiasts is a huge bummer optically….lacking corner to corner shaorness on a7r and on a7 already problems…..and the 35 2.8 is nice but it would be much better if had an f2 aperture….would only slighlty bigger but much better compromise size, dof etc….just my opinion…so fuji xt1 with 23 1.4 has better dof becuase aquivl fullframe 11.4 =2.0….2.0 vs. 2.8 of the sony zeiss 35…
the 70-200 f4 is excellent but large you need a battery grip for good handling… so not much less weight tham dslr….
hopefully the 16-35 f4 is much better optically than its brother the 24-70 f4 zeiss!!
lets pray and hope time will tell..
I agree with most of your points. Sony really needs to develop its FE lens range, but it is early days comparatively.
I bought an A7r for landscape work, as my D800e and Nikkor 16-35mm were beginning to suck the joy out of getting out in the wilderness (due to size and weight). I’ve been using the A7r and 35mm (with tripod) and enjoy it again, but yeah, I’m waiting on the Sony Zeiss 16-35mm and hope it measures up. If it doesn’t, I’m tempted to sell my Nikon gear and get a Leica Tri-Elmar wide… that’s a last resort though.
My guess is we’ll see an 85mm f1.8 soon, and hopefully a 24mm not long afterwards. I’m not overly keen on the digital optical corrections Sony Zeiss employ, but I can live with it given the size and weight advantages.
Yeah, I totally agree. I would get an FE ultra wide, or even the Oly ultra wide 2.8, but they are not out until next year probably. So I am left with no choice but to get the Nikon 14-24 if I need it soon, and I probably will need it soon.
I thought there are already some good native lenses for FE mount from CZ and Sony.
Seems like the best solution is to just buy one of each. The A7r when you need high resolution & the A7s when you want the low light capabilities, or if you want to shoot video. Putting aside 4K, the standard 1080p video on the 7s will be far superior than the 7 or 7r. With both bodies you can still share the lenses & other accessories, so you are not buying 2 complete systems. For those of you complaining about poor battery life, you could always carry an 8000 MW USB power pack and plug it into the side of the camera. Should last you 3 or 4 days, if you don’t mind carrying a battery the size of a cigarette pack. There is no free lunch. Want lots of battery, then you give up size & weight. Due to the fact that Sony designed these cameras for people who want a lightweight system, then those who want to continue carrying around boat anchors, should just move along and buy the older technology from the big two. No need to complain here if you don’t like what they are producing. Just don’t buy into the system. I don’t think that any camera company is making build to order cameras, with the options that each individual customers may choose.
for 2500$ camera in 2014 13.2 EV of DR sucks !
I look forward to the release of your camera.
Me too, quite a disappointment for below ISO 800. At ISO 50-100, it’s 1 stop less DR than other popular FF cameras. What a shame for such low resolution. But Sony claims the camera has 15.3 stops DR, hmm….
However, at high ISO such as above 3200, it beats all other FF.
Conclusion: to shoot best looking photo….look for other cameras. To shoot best looking photo at low light, this is the one to look for. hmm…..
There was a firmware update to allow the sensor to capture up to 15.3 stops of dynamic range in RAW sensor output.
One question to ask is under what parameters is this achievable?
The other question is: Did DxOMark use the updated firmware in their testing or was it still running on the original firmware? That could make a difference too.
Sony should have released updated firmware to allow lossless raw for A7R, not this silly thing
Your comment should read more like:
“Sony should have released updated firmware to allow lossless raw for (pick your Sony camera)”
The A7R is not unique to that situation and I feel Sony should allow all of their current cameras the ability to save uncompressed RAW.
As for putting the A7R before the A7S in firmware releases, the A7S was not released yet and the firmware used by the current users was still in a test or beta phase. So, technically, Sony should be putting the A7S first in firmware in order to get it ready for prime time. The A7R is already released and is using stable, production firmware.
Its good to look it from another point of view. DR=8 is OK for most situations, if you got exposure right. A7s pushes this DR up to ISO 50k ! I say Wow!
A7s is NOT for you my friend. There are other cameras for you in the mkt. 🙂
Its DR is the most homogeneous all along the iso range. 13,2 is just the DR at base iso. At iso 800 it becomes better than all the other cameras in DR. It depends on what you need. Max value at base iso or “resistance”. Different needs, different cameras…
The A7s is the highest rank camera in DXO ISO, but the big difference in sensitivity is in video. Since it is not throwing away a bunch of pixels by line skipping, it gains several stops more sensitivity in video compared to other FF bodies. There are other cameras that use all the pixels but so far have much smaller sensors.
“big difference in sensitivity is in video”
Yeah.Df has no video at all 🙂
Crazy. The A7S has the same DR at measured ISO 150,000 as the Panasonic does at measured ISO 7533 …More than 4 stops …INSANE!
Panasonic GH4 that is 😉
easy one nano sensor vs. monster sensor……..4 times larger sensor 4 stops better…….;)….the discussion is starting…;)
No, ceteris paribus, a 4 times larger sensor would perform 2 stops better (logarithmic…). Obviously, ceteris non paribus est.
Sony cheat. Open SNR graph and you’ll see that curve changes slope between 51200 and 102400. In Nikons curves look natural.
Based on DXO tests, A7s DR and Color Depth disadvantage when comparing to high resolution cameras, is at ISO 800 and
below. It begins to quickly outperform all high resolution cameras at
ISOs above 800, and outperforms Nikon Df in Color Depth at ISO3200 and
above as well as Dynamic Range at ISO 800 and above.
It is clearly an all around camera. It is good in low ISO, great in high ISO, does Video like no other in its range and a good performer all around.
My only gripe about it is its egronomics and electronic manual focus on E-Mount lenses.
Agree with you. In daylight its not much better than Df, but in low light — there is no substitute currently. Waiting for a99s, because of my heavy a-mount lens investments.
Love your work Jack, thanks for posting your URL! ~Chris
Oh, such a strong portraits on your site. WOW!
On the ballet shots how much do you find the EVF & zebras help? The dark backgrounds, hot lights and fair-skinned danders must have given you fits until you got it all dialed-in.
Nightmare scenario with the old FM2 & EPN……(:-)
EVF is everything for me. I swich LCD off completely, I pre-proof (effects ON), not post-proof. Stage lighting is so challenging, that EVF is the way to go. WYSIWYG.
Impressive. We have some real photographers on a gear-headed site. Kudos to both Chris and Jack.
I really like the high ISO performance on paper too, but I’d still rather have the Df until the price drops some; however, if I was into video I’d go for it. Or even if it let me record video directly to SD I might too. Picking out a frame to print would be very interesting.
in A7S in-raw noise reduction kicks in higher ISOs…
sick
I like the textured exterior of the Nikon. Would like to see that on the A7.
I agree. The next A7 or A8 should have a more rugged exterior, but I would like it to be similar in size.
A7s is ok.
But put it in a A6000 body and sell it for $800 and I maybe buy one 🙂
Where is the RX10 APS-C camera? I am waiting.
So at high ISO you get about a 1/5th of a stop better lol wow that’s amazing what about all those lies/claims/bs about massive multi-stop gains ?
Shadow noise, up to 2 stops difference. You’re staring at 18% gray graphs, not exactly shadows. Low light shots tend to be mostly limited by shadow noise in general.
Indeed – of all the measures, 18% grey is probably one of the least useful as (except at ultra-high ISO), it’s dominated by shot noise which means you get similar curves for all sensors of a similar size.
The DR is the better indicator, and the A7S certainly gives a considerable advantage once you get north of 6,400 ISO. The DXOmark low light cutoff point is at a relatively high level of DR, which is why the gap with the Nikon is not that significant. But look at the full graph and you see where the lines diverge. The A7S makes sense if you photograph in dim light and/or need excellent video in such conditions. As such, I can see it being very useful in reportage roles.
Yes if you are going to be shooting at 150,000 ISO the DR difference will approach 2 stops .However even at in camera 25600 you have barely a 1 stop difference. Bottom line if you are shooting under 25600 you gain a tiny fraction of a stop.
At a measured ISO 6400 for the Df, the shadow noise is already higher than the A7S at a measured ISO 14k, over a stop and a third difference.
You get cleaner shadows in low light still (for example at ISO 6400 and 12800), but more importantly with less colour shifts towards purple and blue, even compared to the Df and D4S.
All these cameras and never enough money to have them all. Sony should name the next one the A7$.
+1
You love the price of D4s then!
Good good, now give me that sensor in a A-mount body. I would pay for a D4S/1DX competitor if Sony allows the mirror to move away just like a normal SLR when needed. Something like first shot af lock and then move the mirror away or 7fps flapping the mirror with Continuous AF tracking or a single exposure without the mirror (ala MLU).
Hey admin, what have you heard about RX2 or whatever it is called.
You mean an A-mount sensor with a fixed mirror that eats up that increased low light performance again? (Nothing against the SLT concept but it comes with a small low light penalty.)
Maybe SLT with MLU can help… Back then I think the admin at Sonyalpharumors posted that Sony patented a MLU system on SLT (CMIIW).
Why is RAW always written in capital letters? What do the letters R. A. W. stand for?
Is this a peculiarity of sites relating to Sony camera gear because Sony raw files have the extension .ARW? Do fujcanikonasonicympus sites talk of raw or RAW?
People write “RAW” because they want to be seen as more hardcore than they really are.
The format comes from the word “raw” as it is raw data captured from the sensor before processing and some of the data is later dropped when processed to a final image.
Yep, I write JPEG instead of JPG, cause 1 extra letter makes me 25% more hardcore 🙂
Everybody knows that jpg<jpeg<JPG<JPEG 🙂
Come to think of it, JPEG is more hardcore than RAW, cause it has one extra letter.
Wow, looks like Disquis html completely broke. Thats how hardcore I am 🙂
It happens if you put a “less than” symbol in your post. Something to do with removing HTML from posts isn’t working properly.
As a Minolta oldschool hardcore guy, I use JPE 😉
And the camera makers have realised this as well, most if not all also write: RAW.
Many comments are sort of funny.
Blabla disappointing at low ISO, blabla D4s there Df here, blabla, to expensive blabla, 2 stops here – 3 stops there, blabla.
Every camera to date has its advantages and its disadvantages. So shut up and judge whether the special features of it are worth the price tag or not. If too much is missing go for some other gear.
(btw, I am not nagging about the ones constructively discussing here!)
Many are obviously waiting for the “holy-grail’s-camera”; those ones shall wait for the announcements for the Photokina 2026 (from today’s perspective).
Just kidding.
To be honest – if the FE system is supposed to be one’s all-in-one system, it is not possible with one body, and it is currently not possible with native lens range.
More people should just be taking photos IMO.
Why can’t the A7 cameras close their shutter curtain when a lens is removed or when the camera is turned off? That would be a great feature.
If your idea is to limit the dust that gets on the sensor this would not help. Unless you are working in a clean room any dust that gets inside the camera will eventually migrate to the sensor regardless of the shutter state. This also applies to DSLRs where people claim that there is an advantage to changing lenses with the body pointed downward. The dust particles are in the air. Eventually your body, which is not hermetically sealed, will reach equilibrium with the concentration of dust in the room. It’s just physics.
I suppose so. But it would be nice. I just get so afraid of that big piece of silicon sitting out in the open. My old dslr sensors look as good as new and I’ve never had to clean them properly. I doubt my A7 sensor will look so good in so many years.
More to limit water droplets or something in case the wind is blowing. Being able to brush out the throat or blow out the throat without worrying about touching the sensor would be nice.
Well, I don’t wait until the dust concentration has “eventually” reach equilibrium before I put a lens back on. And there are larger particles (incl. droplets) that primarily move downwards (eg, the result of a sneeze) such I prefer to keep lenses and cameras with their openings pointed downwards, except for brief moments. Bookshelfs have more dust on the top side of the shelf than on the bottom side.
The shutter is way more sensitive than the sensor which by the way has a protective piece of glass on top. A fingerprint on the sensor you can remove but a touch on the shutter curtains causing whatever to it, it’s finito…
I didn’t realize this. thank you.
“It easily beats the second bets camera, the Nikon DF (which costs $500 more!). ” It’s somewhat not really true… The A7s has just a 1/6 stop advantage over the DF…
If all you care about is midtones at ISO 3200. Shadows at higher ISO’s, up to about 2 stops.
Well, but mind that A7s has then over 2 stops advantage over previous Sony 12MP semi-pro camera, the APS-C A700 – the quality of A7s at ISO 3200 is better than ISO 800 of good old A700. But if you count ~3EV built in image stabilization of A700, they are both as good. But when it cames to glass you can put on it, A700 wins definitely. You won’t distinguish picture taken with A700 and F/2,8 16-50 APS – C lens from this taken with 24 – 70 FE zoom from A7s. So, buy yourself A700 with 16-50, shoot raw using Steadyshot and lower ISO, then you can show your picture as taken with A7s at higher ISO and have money still in your pocket. Wow!
So, for pure photography, I would keep A700 instead and lot of money for anything else. Of course, the advantage of A7s is video shooting and that rocks…
This thing is getting out of hand and reminds me of the noise / distortion discussion regarding amplifiers in the 70s. This is not say there aren’t some use cases for which good ISO 12800 performance may be helpful but generally speaking photography is about light. Otherwise don’t bother taking off that lens cap…
Comparing the performance oth the two cameras I’d say they are very much on par except for the ISO topic. Since I am more than happy with the ISO 3200 performance of all up-to-date FF models I would focus on such things as usability / lens range etc. before choosing a particular cam.
Yes, it’s reminiscent of the Total Intermodulation Distortion race (with crappy Japanese printed-circuit, FET amps engineered to vanishingly low TIM levels, yet sounding terrible when driving real loads). Then there was the Watts race, analogous to the current Megapixels race, with Bob Carver’s Phase Linear 700 leading the way at 360 unstable watts per channel into 8 ohms, then the creep of professional amps like Crown into home use, all the while sounding terrible, and actually not as powerful in real-world use into many loads, as were some low-powered Classe, Levinson, etc. that put out huge amounts of current into very low impedances, yet had a conservative “watts” rating.
Like in music, where at some point we stopped looking at the equipment and started listening to the music–hence the resurgence of tiny, flea-powered Single-Ended Triode tube amps–in photography we will stop looking at the equipment and start looking at the images again.
But for now, the races are on.
that..that was beautiful!
Well, put this A7s sensor into A900 body and you’ll get REAL CAMERA to compete with Nikon that I would buy.
So this a7s is not a real camera because its sensor is not attached to a huge and heavy body?
lol
Yes. How else would the laymen know that a photographer is professional?
It is not the size but the features. All of the A7 series cameras are a compromise (cRAW output, fps, etc etc). EVF and LCD are not features but basic requirements of an usable live view camera.
doesn’t A7S have the same 5 fps A900 had?
You need lenses to shoot. The choice of Alpha mount lenses is very wide, you have hundreds of them and they are good. Huge body with big grip is necessary when you use 2,8 zoom lenses or any heavier telephoto lens. So for most advanced photographers that use such lenses DSLR is a must. Well, I tried A7, for me it’s very uncomfortable handheld with anything but lightest 2,8/35 lens. Twists the wrist with 3,5-5,6 kit zoom. Keeping A900 with 24-70 F/2,8 zoom needs less effort. A7 consumes much more power than A900 and has much smaller capacity accus, so you need 2-3 loaded accus for the session taken on one A900 accu. Doesn’t have steadyshot, doesn’t have neither craw+smaller jpg output nor zone feature. Last but not least, if you shoot raw, 24 Mpix A7 or A99 aren’t any better in picture quality than A900 in ISO 100-1600 range that is enough for most uses.
@JiM – Sony has us original Alpha mount shooters by the short hairs, if we want to avoid their Pellicle mirrors. If we’re dead-set on avoiding those Pellicle mirrors, we have to change mounts. Either to one of the two other Sony mounts or to another brand. It’s well thought-out on their part. To most, I believe that it appears to be less of an issue to switch to FE and use an adapter for their current glass than to switch to another brand.
Candidly, because I’ve been unsure of Sony’s direction, I’ve been spending my disposable income on other non-sony items. (Apple, plus some Kickstarter companies have benefited from Sony’s vacillation on product strategies)
There was one period of time, I almost was about to buy an E-mount w/ adapter, but then they released their FE line up. Now I am in a waiting period again for buying Sony cameras and lenses. Fortunately, I can still shoot my mediocre photographs with the cameras I already have. One day, I’ll buy a new state-of-the-art, whiz-bang camera and some new lenses to allow me to shoot my mediocre photographs. The only difference will be is that they will be more costly mediocre photographs, with less noise.
Well, it IS important that the mediocrity not be obscured by noise and other artifacts. I bet you’re better than you think.
Fortunately, A900 with pack of nice lenses is an ultimate machine to create perfect quality photographs for big prints in ISO 100-1600 range. The only advantage of A7s over A900 would be that you would have more details on photograph in let’s say >ISO 3200 range. That would be good to have clean ISO 3200 or more for handheld telephoto shooting, but for such use you need big and heavy body to balance long lens.
I would say that a99s is the way to go. Why? EVF gives read advantage over optical, because of effects ON mode. PDAF is till much better AF system and worth of 1/3 stop light loss.
I shoot raw and don’t need neither effects nor live view. OVF of A900 is one of the best in the world, so far I wouldn’t tell any EVF is better, even if it has its advantages.
Oh, effect on doesn’t mean any special effects of smth. I shoot 100% RAW only. Effects on means, that you see real exposure and exp comp directly from EVF. You see under- and overexposures in real time and there is no need to proof from LCD afterwards. You see end result on EVF while shooting.
If you open “color response” tab at dxomark, you’ll see that CFA are more transparent on the A7S (surprisingly, metamerism score is still better than A7R).
High ISO improvement in A7S over A7R is 34%, the same improvement A6000 has over NEX-7. That is, all the improvement is due transparent CFA, not larger pixels.
So DxO values again debunk fallacies of pixelphobes
It actually looses a stop of DR at low ISO compared to the A7R, as low ISO DR is dominated by read noise, there is no advantage with large pixels.
It does, but your reasoning is incorrect.
at low ISO DR limited by ADC/amplifier.
at high ISO DR limited by read noise — and here larger pixel cameras have marginally better results.
As pixel area increases, read noise does, too, but it grows slower than pixel area.
By read noise I’m including everything that happens after light hits the snesor, including ADC (Sony Exmor sensors have such great DR because of steps taken to eliminate noise on the way to the ADC).
High ISO DR is limited by photon noise, there simply isn’t enough light being received in the shadows hence the random nature of photons destroys any detail.
Lumping ADC and readout in one number might be a good thing for shooting with one camera, but useless to compare two markedly different cameras.
Sorry but experimental data does not support your conclusion, Dmitry. Please compare raw data, say, A7r vs. A7s. You will see that the brightness (calculated as a weighted sum of R, G, B, G2 components, same weighting formula for both cameras – you can use HSB readings in Photoshop) is very close.
Why would I measure when DxO did it already? I trust them more than you (and btw they have at least published color matrices)
Are you sure you have are interpreting DxO results accurately?
BTW, regarding the other test which you hadn’t done yourself but predicted a result, you didn’t speak when the result was the opposite.
Please be more specific. What is that “other test”?
Now, I know that dxomark sometimes doesn’t really show what will happen in real life but 13 DR at Low ISO (<ISO800) is truly underwhelming. Plus, no internal recording, bad ergonomics, usual AF 'FE related' issues and other disadvantages lets me think that Sony had done something wrong with this camera. I mean, surely I thought this was a very nice niche camera, but I do prefer either a conventional high MP count FF or something cheaper but 4K ready to shoot videos (Sony FDR-AX100 and heck even the Pana FZ1000).
Wait, what’s wrong with SAR’s/Disqus’ HTML?
If you have a “<" character in your post, it will break.
Ugh, that was embarrassing. Thanks for clearing that for me!
Well I was hoping for 15 stops of DR!
I’m still at a loss… the camera doesn’t overwhelm the competition at still shots. Its main selling point just seems to be as a video camera, which is fine… except if that is the case why in the hell don’t they package the sensor in a camcorder style case, give it some needed camcorder items like XLR connections and of course the ability to record in the camera itself not some add-on third party box.
Overall score for Sony was 87 and Nikon 89. So not sure of heading of article. But if low light is the sole factor then Sony is a bit better but at a cost. Mega pixels. 16 in FF is the lowest I would go for lower is OK for smaller cameras – more lens selection with the Nikon but larger body. I think that the A7/R/S are to small for longer faster zooms like the 70-200.
When I said some days ago that this sensor starts noising at 3200 Iso, same as other cameras and that all depends on good software to reduce noise, some pantpissers thought to give me a lesson and insult me. But all that leaves me cold, simply because the truth will emerge by itself with the first stills testing of this camera. Now, the pictures are online, and even at noise reduction 0, we can see very clear that Sony applies noise reduction, the colored spots are softened and look blotchy, and this picture at 3200 Iso, NR at 0, will tell us the truth. I leave you all see what follows after that by yourself, at higher sensitivities.
http://www.imaging-resource.com/PRODS/sony-a7s/FULLRES/A7ShSLI003200NR0.HTM
Look at the shady areas, and the bottle tags, the shades behind the wool,
now compare A7R
http://www.imaging-resource.com/camera-reviews/sony/a7r/FULLRES/AA7RhSLI03200NR0.HTM
and A7
http://www.imaging-resource.com/camera-reviews/sony/a7/FULLRES/AA7hSLI03200NR0.HTM
Now, look and see where you find a better resolution. At the muscat vine bottle the tiles are best at A7R, A7 follows, A7S is a bit less good. But, surprisingly, A7R and A7 fail in the deep red colored tissue, where A7S has a real advantage on the 2 first.
So, make up your mind, at 3200 ISO, the 3 cameras have noise, A7S look here better, but not much. Now, go to the site of Steve Huff and look his pictures, all of the high ISO shots are in black and white, and everything higher than 8000 ISO looks in a way I would really not like to see it in color. I do not comment further, take a nose full and do that to yourself. If ever I decided to buy A7S, the reason would certainly not be to be able to shoot clean shots at 25000 iso in color. 3 stops they say, they really need glasses.
May I ask you, are you looking at conversions when you say “Sony applies noise reduction”? May it be that “colored spots are softened and look blotchy” is because of the converter applying noise reduction?