Sony Artisan Brian Matiash posted that Sony A7II hands-on video and confirmed in it that some third-party lenses will “only” get the 3 axis stabilization. Would be nice to know from Sony why those lenses cannot get the full 5 axis stabilization…
More A7II news:
The A7II shipment date also changed from Dec 9 to Dec 11 in USA (Amazon, BHphoto, Adorama, SonyStore).
New Japanese tests at Yodobashi and Kasyapa. Norwegian first impression by Toralf Sandåker.
If you find more test feel free to post them all on our forum thread: sonyalphaforum.com/topic/135-sony-a7ii-steadyshot-test-la-ea2-sal70400g-ssm-ii/.
USA/Canada preorders:
Sony Alpha a7II (Body Only) for $1,698 at Amazon, BHphoto, Adorama, SonyStore and Sony Store Canada.
Sony Alpha a7II with FE 28-70mm OSS Lens for $1,998 at Amazon, BHphoto, Adorama, SonyStore.
Sony 70-300mm f/4.5-5.6 G SSM II Lens for $1,148 at BHphoto.
EU preorders:
Sony Alpha a7II (Body Only) for 1.799 Euro at Sony Germany, Sony UK, Jessops UK, Sony France, Sony Italy, Sony Spain, Sony Austria, Sony Switzerland, Sony Netherland, Sony Belgium, Sony Portugal, Sony Sweden, Sony Finland, Sony Norway.
Sony Alpha a7II with FE 28-70mm OSS Lens for 2.099 Euro at Amazon Germany, Wex UK, Jessops UK, Verkkokauppa.
Asian/Australia preordersNote:
Sony A7II at Sony Store Australia (Click here) and Amazon Japan (Click here).
136 Responses
It looks like Sony Marketing took these 2 axes of stabilization.
It’s a technical issue of distance encoding, not a marketing one.
Are you sure? because focal length is set in menu, and distance can be measured by AF pixels, or by checking contrast.
no it cannot, how should the camera know how far a subject is away without the data from the lens?
The E-mount lenses show the focus distance on the screen if you manual focus. You will never get this information with let’s say a Leica lens without electronic contacts so that’s the information missing for the last 2 axis to work properly.
I’m sure it’s no marketing conspiracy thing.
Contrast AF will not tell the camera how far away something is. It just essentially does what it says on the can. Maximises the contrast. A 50mm lens could be focussed on something 10cm or 10m distant, and the camera won’t know unless the lens tells it.
It may be possible with PAF to work out something like this using the phase detection difference across the baseline (I’d have to think abut that quite hard though), but as third party lenses won’t work with PAF then it’s not really relevant.
right, contrast AF does not know anything about distance. (can it be used together with phase AF for confirming which PAF points are in focus?)
it can and is on the hybrid designs, but it seems that for it to work at all well there has to be a very tight communication between the lens and body. It’s notable that MFT hybrid AF doesn’t work with the original four-thirds lenses, just newer MFT ones. Also, I seem to recall that some E-Mount lenses required firmware updates.
With new lenses all bets are off of course as heaven knows what information exchanges are going on.
Gary Fong put up a comparison between the A7 and the A7II. AF seems improved quite a lot…
https://www.youtube.com/watch?v=V7vhTIDY_RQ
How? It couldn’t even follow a slow moving train…..He had to praise the a6000 for the ability to focus better……
To me the focus in this video looks not good.. but maybe the experts here can comment on this 😉
Well to be fair it was better but just not an awful lot.
Pitch and yaw needs distance to object. That makes it impossible with old not info connected lenses! 3 axis is not bad for older non Sony lenses.
Why should that be necessary? Both pitch and yaw are about angle, not distance. I haven’t heard the Olympus E-M5/E-M1 having any restrictions to their 5-axis stabilizer with manual lenses, either. (And by the way, I think no existing camera is able to deliver really exact distance information with all automatic lenses, if only because many have too large a depth of field to give any conclusive distance information just by reading where the AF locked on.)
I can’t quite wrap my head around it either actually. What axes cannot be compensated without knowing focus distance? But regarding your second question, I should think IBIS along those 2 axes only needs an approximate focus distance to work, doesn’t have to be super exact.
Pitch and yaw (and roll) are independent of distance, it is vertical and horizontal translation that require distance information (due to parallax, close objects appear to move more than distant objects).
As somebody else pointed out, for correcting some types of camera movement, the IBIS system has to know how far away the lens is focussed. That means you need an electronic link to the lens which provides this information.
It would be interesting to know if any A-mount lenses will have full stabilisation. Precise distance encoding was introduced into A-Mount lenses when digital sensors were introduced, largely to provide a for a new automatic flash exposure system (ADI) as the old TTL mode (measuring reflection off the film surface) doesn’t work with digital sensors as the surfaces are inherently reflective. (There seems to be some doubt to what extent pre D-series lenses had a crude form of distance encoding).
Which types, and why? And why should a lens need to be focused at all for stabilization to work? And as mentioned in reply to that someone, I don’t think there is a way for real-world cameras and lenses to reliably determine distance at all by evaluating an auto-focusing result.
I believe it’s the linear movement along the A&Y axis which requires the distance,. Of course what this means is that it can only compensate for the area where the lens is focused, but as that’s where the sharp image is to be found, that’s the part where it matters.
Of course, the camera needn’t actually be “in focus”, but the lens will always be focused at a given distance. I think the correction required is larger the closer the subject and it gets increasingly less important the further away it is. For many circumstances it won’t be at all significant. Pitch & yaw matter far more.
Also, like focus, whilst there might only be one exact distance which is truly corrected, there’s sufficient latitude that it covers a wide range of distances, especially far away. It will be for macro subjects where it matters most.
(I’m doing most of this in my head at the moment – I’ll need to get the pencil and paper out to make sure).
So when I move my head sidewards and vertically, of course the distant objects will appear to shift much less than the foreground objects. But the IBIS would be if my retina shifted sidewards and vertically in the reverse direction no??
It’s a bit tricky to talk about the retina shifting sideways as we actually move the entire eyeball as one to focus on something. There isn’t any side-to-side movement of the retina. It just moves around with the eyeball as we redirect our gaze.
However, if we were to be able to do what you suggest, and could slide the retina sideways, we would, indeed, have to move it further for the closer object than the one farther away for any given movement of the head in the up-down/side-to-side plane. It’s also exactly the same whether it is up-down or side-to-side (or any combination) as it’s all symmetrical around the lens axis.
nb. as an aside, this sort of thing leads thoughts into the mirror paradox whereby reflections are apparently transposed left to right yet top/down is not. That’s actually a perception issue and a surprisingly difficult one to understand, although there are lots of misleading explanations about.
I would disagree on one point and agree on another:
The AMOUNT of shift needed to correct for X/Y movement DOES NOT depend on the focus distance. That is, if I were to move my camera body 1mm to left, the sensor would need to move 1mm to the right relative to the body to correct.
However, to MEASURE the amount of shift needed MAY require knowing the focus distance as the amount of movement observed on the sensor WILL depend on the distance to the reference object. So, if I had an accurate way to measure the shift without using readings off of the sensor, such as an accelerometer, I may not require knowing the distance to a reference object, but if I am looking at how much something moves on the sensor, I need to know the distance to the object.
What would further lend credence to 5-axis requiring knowing the focus distance is that the Sony lenses that do not support 5-axis are the same ones that do not support ADI, according to the interview with Sony Europe posted here (they did not mention that it does not work with the 135mm STF in the summary here, but the others are all Sony lenses without ADI.
It’s actually quite easy to see that the amount the sensor has to move depends on distance. Simply sketch a plan view of a pinhole camera on paper. Now draw a straight line from a point on the image plane out through the pinhole into what would be the field of view. Draw two points along that line to represent two objects that would appear to be lined up at different distances from the camera.
Now imagine we move the camera (say) 1 cm to the left. It’s a bit of a pain to redraw the camera, so it’s much easier to draw two new pencil points 1 cm to the right or the original and the same distance from the image plane (which is logically equivalent, just easier to draw). Now draw straight lines from the new pencil positions through the pinhole to the sensor surface. Note that they are no longer lined up, and each will hit the sensor at a different location. Note also that the new image point further away object will have “moved” less than that for the nearer one.
There’s no particular significance to using a pinhole camera. It’s just easier to see the principle, and you don’t need to worry about what’s actually in focus. You can see that it’s not possible to move the sensor in such a way as to compensate for both. With a pinhole camera, everything is in focus, so sensor displacement wouldn’t work well as the foreground appears to move much less than the distance. However, with a real camera, then only so much is in good focus and the smudging of those out of the DoF area matters less. (Also, the geometry is such that for things a long way distant, the effect is insignificant anyway as is the differential effect). However, some people claim to dislike what they believe to be artifacts of lateral image stabilisation on out of focus areas, and this gives it some credence. Also, it looks like it will be more of a problem with macro photography, especially where there’s a significant DoF (admittedly a difficult thing to achieve with macro).
Of course this is all just a consequence of perspective, but it’s much easier to draw out than explain.
nb. the focal length of a pinhole camera is just the distance from the pinhole to the image plane. You can therefore easily use the same technique and modify it a bit to see the impact of focal length and pitch/yaw. Indeed you can cut out a plan view of a pinhole camera and really move it around to examine the effects.
http://en.wikipedia.org/wiki/Pinhole_camera
Right, take your example, and after you have moved the dots, redraw the camera so that the 2 points are now back in line. This will be 1 cm to the right of where it was. Now, repeat this whole thing again, but select two different points further away on the same line. How much will you need to move the camera to bring the points back in line? It will always be the same distance as how far you shifted the points initially.
So, once you know how far the camera has shifted, to get the image back to original, you need move it the same distance in the other direction.
However, to calculate how far the camera shifted based on how far a reference point moves on the sensor you need to know how far that point is from the cardinal point of the lens.
In particular, the physical shift of the sensor (call it x), is given by:
x = (df/di) * y
where df is the focus distance (or the distance between the cardinal point and the reference point), di is the image distance (distance between the cardinal point and the image plane) and y is the observed shift on the sensor.
di can be derived from the focal length (assuming we are modeling things as a thin simple lens). y is determined looking at how much our reference point shifts on the sensor. So, what we still need is the focus distance(or distance to our reference point). This is why the distance information is needed, but once we calculate the physical shift of the sensor, the correction is to move the same distance in the opposite direction.
Sounds right. I’ve been doing this all in my head as I’m too lazy to go and find where I’ve put the ruler and paper to plot it all out.
Probably somebody ought to write a blog item about this with some diagrams for all the degrees of freedom. The number of times I’ve had o argue the case for focal length information being required is amazing, despite all the evidence.
I think the pinhole camera approach makes it fairly easy to appreciate. The pitch/yaw bit can simply be represented by moving the two pencil points on an arc, although unless the centre of the arc (and hence the yaw/pitch axis) is centred on the pinhole, it becomes a bit complex as you get a mixture of lateral and rotational shifts.
It might be possible to produce a proper animation of this. Something a bit better than Sony’s promotional stuff.
Wrong. If you move the camera 1 mm to the left, you will have to move the sensor to the LEFT inside the camera, not to the right.
And the required movement will not be 1 mm. It will be roughly 1 mm * [focal length] / [subject distance from lens]. That is why you need to know the subject distance.
the IBIS is your whole head moving and following you subject. So your eyes lock on the subject and your head knows the distance and can compensate the movement. Without focussing on the subject first your head would be shaking like you’re drunk. That’s why Sony deactivated those axis.
No. The IBIS system is compensating for modest of the whole camera (plus lens) so it is the same as moving your while head.
Just think about what happens when you move your head sideways: the foreground objects will appear to shift much more than objects in the very far background. Like Steve said below, IBIS has to choose one part of the image to compensate for. Electronically connected lenses don’t feed extremely precise focus distance to cameras, but it only has to be approximate.
That is not correct, the Olympus IBIS system offers 5 axis on all lenses regardless of an electronic link. Even on manual focus lenses. So it can be done. Sony for whatever reason have not chosen to offer this.
I don’t think they do, since this is not actually possible. I think Olympus simply don’t mention it. Translation is generally a small component of camera shake anyway.
How do you know it is not possible… you don’t. You could soon tell the difference if it was not 5 axis. Remember Sony are not using Olympus’s 5 axis and that could be the problem. It is just a different system.
It’s not possible as the movement of an object across the sensor plane due to translation of the camera is dependent on subject distance – which is provided to the camera by the autofocus system. If you have a completely manual lens this information doesn’t exist, therefor you cannot compensate for translation.
5 axis = pitch, yaw, roll, horizontal movement, vertical movement
Horizontal and vertical movement only matter when you have objects very close to the camera, so no, you wouldn’t immediately notice it’s absence.
Almost all lens based image stabilisation systems are 2 axis only, and they’re still good for 3 stops.
Just feel the need to order train tickets when I see this video. Will be nice to compare this camera with the Nikon 750, with all the old Nikon lenses I’m using right now not sure my third body wil be a A7II.
Why is the shutter noise so loud?????
Mirrorless, so shutter is open as long as you don’t take a photo and has to close first before you can shoot.
With electronic first curtain shutter (EFCS), the shutter doesn’t have to close first (not an option on the A7R of course, but it is on the A7 II).
Yep, love it on both my A6000 and A7S, but I wish sony would put a silent option on all new camera’s.
I’m confused. The A7S has a silent shutter, but the A6000 only has EFCS as far as I know.
The problem with putting a silent shutter onto the higher resolution sensors is that it isn’t currently possible to read and digitise the data fast enough. On current CMOS sensors (apart from specialist ones with a global shutter), this works using an electronic analogue of a focal plain shutter by running an EFCS followed up an electronic second curtain which essentially is just a traveling strip reading each pixels line by line. It can only pass down the sensor at a certain speed, which makes it vulnerable to rolling-shutter effects (e.g. leaning verticals on a moving objects). Mechanical shutters have this to, but they travel at 10-20 times the speed. The A7S only manages a tolerable result because it has only 12MP and also compromises by reducing the bit depth.
EFCS works a lot faster than an electronic second curtain as it only has to clear charge levels, not read and digitise them. Even so, the A7R doesn’t have EFCS, probably because at 36MP it has too many pixels even for that.
Hopefully, one day a truly global shutter will appear on CMOS which doesn’t compromise quality. That will solve a lot of problems.
Yawn! this is such a non-issue for all but the most noise sensitive applications. Internet amplification effect runs rampant on just about any whim on SAR.
It only seems loud to trolls with sensor and IBIS envy.
The great thing about lens adaptability is that you can own and select from several outstanding e.g. 50mm primes that each render differently. No longer does choice of body limit the range of rendering available to a photographer (though new incompatibility across mirrorless mounts is already eroding that freedom). Other factors like size/weight, field curvature, tilt/shift, prime/zoom, etc. are also common reasons to own multiple lenses covering the same focal length. It is a shame Sony don’t have enough confidence in their own lenses to believe most photographers will choose Sony lenses unless Sony witholds or limits capabilities when non-Sony lenses are mounted. A fully open approach to adapted lenses, supporting or enabling others to support everything from autofocus to stabilization, would signal a much stronger commitment to excellence in lenses, and give me a lot more confidence in Sony’s long term commitment to the whole system. Selecting from the widest possible range of lenses helps me get my ideal result more easily and more often, which is the best way Sony could earn my loyalty. That is, after all, why I spent a relative fortune on an ILC and cabinet full of lenses instead of a top quality superzoom…..
For the last two axxis the camera needs to know the distance to the subject, since legacy glass does not provide any distance information to the camera it can not calculate this distance.
In my opinion the Sony approach is very open for legacy glass.
Show me the system does comes even close to what we have here!
Canikon: Forgett it!
Other mirrorless: No Full Frame.
A7 series: Your only choise!
That is incorrect. Olympus allows 5 axis even on legacy glass as it does not need to know the distance.
Yes, Olympus has somehow changed the laws of physics. Idiot.
My guess is Olympus also only do three axis correction with legacy glass but haven’t extensively documented this as most users wouldn’t be able to tell the difference.
No they are simply using their own system. Have you actually tried using an Olympus with the IBIS switched on with a legacy lens… I am guessing that is a no. Has anyone every reported it is only 3 axis with legacy lenses… no again. Does Olympus say it is 5 axis with all lenses… yes. Try using it first of all before saying it is not possible. Try reading up about something before saying if it is possible or not first as well. That way you will not make yourself look so ill-informed in future.
For some reason Sony has chosen to do it differently and perhaps due to patents they are unable to offer 5 axis on all lenses. That is just a guess but whatever it is they are not using the same system of 5 axis as Olympus or they are holding the feature back permanently or maybe for a future A7 III.
So buy a Olympus and be happy..
I have also been saying Sony should provide more compatability with other lenses such as Canikon.
If they are unable to allow rapid autofocus with slr/slt lenses then the laea4 should be upgeaded to be as fast as the most recent A mount cameras, and a similar adapter should be available for non Sony lenses such as Canikon.
Sony will benefit by having greater access to the vast majority of consumers that are currently Canikon lenslocked and this will greatly increase sales…especially when the A9 comes out.
More importantly the consumer will benefit by being able to use the equipment they actually want to use rather than just the equipment Sony wants them to buy.
In a famouse old movie…Miracle on 34th Street…against Macy’s wishes a store Santa tells shoppers where to buy products Macy’s doesn’t carry.
At first management is furious but then customers are so happy with the new policy they are buying more than ever. Soon all the major stores in NY are telling customers where to find the products they want regardless of who sells them.
This is how cameras should be…all lenses should work fully with all cameras.
Arca Swiss made thier camera mount non proprietary and available to everyone today it is considered the world standard.
Sony could lead the industry in making full compatability a standard and they will benefit by doing so.
Some people comment that they wont because it is in Sony’s best interest to keep everything proprietary…besides the fact that I dont think that is really true , I also don’t care.
I will not buy a camera or lens because it is in Sony’s best interest to sell it to me, I will buy the product it is in my best interest to buy.
Great, now what? they should stop developing their own lenses and work on making whatever canikon work on top of being compatible with minolta and stabilising old manual lenses? If you want to use old DSLR lenses, don’t buy a mirrorless from Sony.
I’m buying it because I ant to be rid of the old heavy canikon stuff and get me some lightweight Zeiss goodness, I want them to put all their money into making new FE lenses, not waste more resources. How about Canikon offering a upgrade service for their products to make it compatible with Sony? They are making money from these lenses, not Sony, why you expect Sony to care?
If you want native e mount lenses you will continue to buy them.
If you want Canon lenses and they do not work on Sony camera…you will not buy Sony camera or Sony lenses.
If you buy Sony camera but dont want Sony lenses you will buy Sigma or Tamron lenses not Sony lenses.
If you buy Sony camera that that works with your Canon lenses…you may also buy Sony lenses too.
And theres a lot more of you than Sony users.
Get it yet?
Oh I just sold all my Canon lenses & camera’s so I guess, you want a sony camera with canon lenses. Off course I ‘get it’ its a childish fantasy, its just not very reasonable to expect sony to develop for canon if they obviously don’t have the resources to produces a lot of new lenses already.
It makes total sense to sell your customers what they want to buy.
It makes far less sense to try to force customers to buy what you want to sell them when they are free to buy from your competition instead.
And what customers want to buy most is fast FE lenses, so they put their efforts in developing those, instead of making money for Canon, yeah I know totally weird and selfish of Sony. go complain to canikon you can’t use their lenses on sony camera’s see if they think that makes good business sense.
Making the camera compatable does not prevent development of lenses.
Just how big a percent of Sony’s resources do you think will be tied up by making a camera compatable or making new adaptors…a rookie Sony engineer apprentice should be able to whip something up during his lunch break.
Yes, we all know writing a bit of software is dead easy, anyone can do it, even a sony intern, in fact, why don’t you do it? You seem to know how it all works, go talk to Metabones and tell them a rookie sony engineer could have made their adapters work perfectly well and they are just crap by not following your imagined lens utopia.
How about this, go to the sony website, look for their phone number and CALL them, repeating the same idea on here is not going to change a thing.
Sony does not have to be the one making the adapter, any 3rd party can do it, just as firms like Sigma and Tamron make lenses that work with Sony cameras.
I think the reason no one has is that Canon and Nikon would sue anyone who does.
The that Canikon might try to stop anyone from adapting thier lenses to Sony is a real possibility.
The idea that Sony should not want this to happen….as some seem to think….is ludicrous.
A video on a camera you have had for 1.5 hours? Really? How can this be considered valid by any stretch of the imagination. Well, I guess at least it isn’t as bad a Tony Northrup’s 20 minute video with giggling wife on a camera he has never even seen before and doesn’t even know what type of memory card it uses. The race to be the first to post videos about something they have no experience with has become absurd.
Very well said!
Tony Northrup has a wife he has never seen before for 20 minutes?!?
I think the giggling through the video was very unprofessional, she was reacting to bloggers making a joke about the black lens being longer than the white one. Beavis and Butthead style humor has it’s place but not in a camera review.
Sony is getting better and I like how innovative they have been lately. However, I don’t know why or like that they compress the raw files. Also, I would prefer an A7r ii, if they could resolve the shutter vibration issues.
Hopefully Sony stops restricting the raw files to compressed raw files, as some of us do not want that limitation. The D800E/D810/D 600/D610 have better image quality just because of this.
This stuff is so academic and only repeated by the clueless! I could easily take your Nikon (uncompressed RAW) put on Nikon’s best 50mm and shoot those star trails wide-open and get nothing but mush in 1/3rd of the frame because the Nikon 50’s are pretty lousy. Now, I could repeat said exercise on an A7 (compressed RAW) with Sony’s 55mm and get an image WIDE-OPEN that is extremely sharp RIGHT ACROSS the frame. See where this logic is going? ITS GOING NOWHERE.
No, 50/1.8g is very sharp a little stopped down (and much much cheaper). The Sony 55 is a really good lens though, one I would purchase immediately with any A7-body.
I agree with Jen. Nikon’s 50mm lenses are really terrible. With Sony’s 50mm F1.4 Zeiss Planar for A-Mount, you don’t have to stop down at all to get a sharp image, and I imagjne the 55mm FE F1.8 is similar in that regard.
Being almost exclusively a 50mm shooter these days (and an OVF addict), I sometimes consider that I might one day have to switch to Nikon or Canon, but I would never buy the Nikon 50mm F1.4 G or the 50mm F1.8 G. And the Canon 50mm F1.4 is nothing special either. (I know cause the second shooters I hire use all of these lenses and I have to do post production on their work). I haven’t seen first hand what the Canon 50mm F1.2L can do wide open or the Nikon 58mm F1.4 G, but those would be the only 50’s that I’d consider on those systems I ever switched.
For now, I’ll just keep shooting my a900 and loving the results. Oh yeah, and the a900 can shoot either RAW or cRAW, so even if I was going to shoot star trails and bump the exposure up 2 stops, I guess if have the option to still get a good shot by just switching back to RAW. For the record, I always shoot in cRAW as I’ve tested both modes, and can’t see any difference upon close inspection in lightroom.
I have the 50/1.8g and Sigma Art 50 1.4. The Nikon is very sharp on my D610, but not able to match the resolution of the D810. It’s very cheap and very lightweight on the other hand, a perfect travel lens. The Sigma is in another ballpark. But to call the 1.8G terrible means you didn’t use it.
Ok. Let me rephrase. The Nikon 50mm F1.4 G is terrible wide open (reddish purple hazy edges around what should be sharp, even in low contrast light, and no I’m not talking about the purple/green fringing in the front focused regions and bak focused regions that also shows up and which is also terrible with that lens), but, having said that, the Nikon 50mm F1.4 G does have quite lovely bokeh when shot wide open.
The Nikon 50mm F1.8 G (which I have indeed also worked on images shot with it on a D600) is, in fact, slightly sharper than the Nikon 50mm F1.4 G when shot wide open, but it has much less pleasing bokeh, and therefore is not something I would consider for professional portraiture for my own work. At least not at the level that I expect from my 50mm Zeiss Planar F1.4 for A-Mount.
Perhaps the Sigma Art 50mm F1.4 is an excellent lens, but I’ve never worked with files that were shot with it. I know a lot of documentary wedding photographers shy away from off-brand lenses due to less reliable autofocus. We work fast, and anything that’s gonna cause us to miss focus more often is, well, to be a avoided.
The autofocus of the Sigma is quite good and accurate. I often use it at 1.4 in dim light without issues. Took me a little time to adjust it with the usb dock, but after that its tack sharp. It’s considered second to the Zeiss Otus only. The 50/1.8g is a lens I don’t use for portraits, anyway. I prefer 85 and up for that.
Is it on Nikon’s very own “recommended for their SONY 36MP sensor D800E list” — if it isn’t, why isn’t it?
As I said, performance near the borders is not on par with the Sigma ART or Sony 55. I bet the 55/1.8 will even be great on upcoming 50MP Sony sensors. I use it for traveling due to the weight. But, I need to buy a Sony for Christmas anyway. My wife urged me to sell the OMD now that IBIS is in the A7ii. As she desires a 135mm we now add the 135/1.8 Sony-Zeiss + LA-EA4+A7ii to our mix, since the 135/2 DC is too soft wide open (1990’s lens construction). Can use it for handheld macro, too, using the 150mm/2.8 Sigma via Novoflex adapter. Will be my first Sony after using a walkman back in my youth.
It would be hard to focus that 55 on stars though.
A lot of people just react against the idea of RAW compression in principle, not because it’s got any real significance. There’s really only one circumstance where it can be shown, and that’s a fairly extreme one and is of little practical importance. From what I recall early A-Mount digital cameras used to have the option for compresses and uncompressed RAW, but I think people tended to use the latter if only because flash cards were so expensive in those days,
First some of the reason why it’s possible to use fewer bits without losing real image information. That is for areas of broadly similar illumination, there is never actually anything like 14 bits of information. In the dark areas it’s fairly obvious why. The most significant bits are all zeroes, so why hold them? However, it’s less obvious why there are fewer bits of information in the brighter areas. Well, the answer to that is one term. Shot noise. A typical photosite might saturate at about 16,000 electrons (the actual number is mostly dependent on photosite area). To represent all the numbers between 1 and 16,000 electrons on a linear scale you need 14 bits. However, not so fast. Not all of those electrons represent real image data. They are “noise”, and what actually matters is the signal to noise ration (SNR). As it happens, the noise in bright areas is dominated by “shot noise” due to the random arrival rate of photons. Shot noise varies as to the square root of intensity, so at a 16,000 saturation level, the “noise floor” will be about 126 electrons (it will be slightly higher due to read noise as well, but that’s relatively insignificant). Of course that actually means that of those 14 bits, 7 bits are just noise. In theory, the actual signal intensity is only in the most significant 7 bits. Now there are good reasons why it’s best to digitise below the “noise floor” as you are in danger of posterisation. However, 2 bits are plenty for that purpose. With less bright areas even fewer bits are required.
So, in principle, you never need more than about 9 bits anywhere (usually a rather less). The problem is which bits as the scene varies in brightness across the picture. The answer to that is to use an adaptive compression system that works in very small blocks. The only time this could become an issue is where there are tiny parts of otherwise bright areas in these tiny blocks that are “boosted” to bring out deep shadow. Then there have been a few examples of artifacts. However, for the vast majority of circumstances, this is insignificant.
Also, it’s further up the ISO scale, then this is of even less importance as the SNR degrades due to much lower signal levels. Boost the ISO by 4 stops and you lose 2 bits of levels in the bright area (that shot noise square root function).
Nb this form of compression first appeared with Nikon as far as I’m aware.
Amen
Good analysis. Additionally, real scenes, although having more than 10 stops of DR totally, often show about 4-8 stops of DR for nearby positions, only. I read somewhere (can’t remember the source), that the problem which should be seen in regions of abrupt changes in DR across 32 pixels (1/10 of an inch in fulll size prints) is mitigated by LR doing interpolation.
Thanks. I simplified how the compression system worked (it’s rather more adaptive than I hint at), but the point I wanted to get across is that there is never actually 14 bits of real data in any one narrow area. A lot of it is just noise.
As you say, I can image that interpolation does reduce the impact of sudden jumps (which I only think happen in otherwise bright areas where a lot of “boost” is applied, and then only one low ISO shots.
However, it remains something of a mystery why Sony don’t have a non-compressed option. Maybe it’s data rates and that might affect buffer capacity, frame rates or whatever. Of course it would be a user choice, but then doubt it would still be a criticism.
Still, I think this is another one of those issues that is of little real world importance which just gets used as a stick from time-to-time and the easiest way might just be to acquiesce.
“Nikons 50’s are pretty lousy”
You just lost all credibility there.
Unless they’ve released something when I wasn’t looking, the 55/1.8 beats the 50/1.8g and 50/1.4g hands down.
If so just use a Sigma Art lens instead on the Nikon.
Let me think, 815g lens vs 281g lens on a 416g camera…
Andrew, they’ll find any reason to bang a round peg into a square hole rather than admit, for them, the A7 & Sony FE glass is perfectly adequate!
I can’t understand your problem. If you have an A7, go with 55/1.8 (great lens). If you have D750 or D810, go with Nikon 1.8G or Sigma Art. If the 55/1.8 wouldn’t be better at six (!) times the price of the 50/1.8G, they would have a problem. In the center, differences are quite small, at the borders, the Sony leads until 2.8 where the cheap 1.8G lens is almost on par (36MP sensor, picture shows it wide open). Looks much better on the 24MP sensor, by the way. For me a great bargain lens, the Sigma is for the heavy lifting.
It better should for the price. I got the 50/1.8g for 150 Euros new, the 55/1.8 sells for almost 1000 Euros here.
Except I love the 50/1.8 bokeh every single day of the year of the clinical 55.
And actually I don’t think any other 1.2 beats the Nikkor AIS (thought some other 1.2er are different but still same level) in both beauty and sharpness.
What a load of old rubbish………. It’s soft & mushy and won’t resolve your superior Nikon file………. gee-whiz. You could put that 50/1.8 on a Sony and save yourself the embarrassment of another hernia.
And once again, the trolls are so busy trolling they don’t read their spoon-fed “talking points.” Shame on you!
http://nikonrumors.com/2013/04/29/nikon-published-an-updated-list-of-recommended-lenses-for-the-d800e-camera.aspx/
And just in case you can’t get to the bottom of that list for all the tears in your eyes, that’s ONLY 23 lenses from Nikon in over 60 years of development that they NIKON recommend for the SONY SENSOR they use in the D800E. Makes Sony’s FE lens line-up look pretty impressive after what, 12 months……. Just Sayin’
But the Nikon raw file will capture all the mush in exquisite detail, satisfying the most demanding photophile.
Or blogger that never picks up a camera. LOL
“The D800E/D810/D 600/D610 have better image quality just because of this.”
I’ve never seen a comparison the support your statement in real live. Do you have a link?
The only picutures I’ve seen on the web regarding compression are from long exposure star trails that have been pushed +2 stops in post, so you could see some disturbing pixels in areas that have been black before.
Believe me, for 99% of us this compression has the benefits of much smaller file sizes, for 1% it might be relevant and visible in their results, but only if they are intentionally searching for flaws.
This RAW compression topic is way too overrated from what I’ve seen.
Like shutter noise, simply a troll taking point that keeps getting repeated over and over until people actually think it’s a legitimate problem. Amid the hustle and bustle of a busy city street the momentary click of a shutter is cause for concern for a street photographer, really. Don’t take these concern trolls seriously, they are trolls after all.
Personally I like the click to be a real CLICK, not some sissy fart.
Completely agreed.
Not that I have good technical knowledge on the issue but it worries me as I frequently pull and push exposures (i.e. adjust dynamic range) on PS. Do you think my worries are substantiated?
I’ve found the Sony raw files to have a very wide latitude of exposure adjustment in Camera Raw (Photoshop) or Lightroom. The recovery of shadow areas and highlights is excellent, with little added noise or other artifacts.
I push my shadows all the time and haven’t seen any problems. But maybe it’s because I don magnify to 400% after I did so to find 10 out of the 36.0000.000 pixel that could eventually look better if it was not compressed 😉
They Might do that in 6 more E-Mount body releases, You know, 12 months from now.
Maybe the 7th. Cant have you having a camera that totally makes you happy/Makes sense. You wont buy the new one that comes out in 1 month, Just to realize something else is handicapped, and do it all over again 2 months later.
All the better… much better than Fuji X or Oly
Maybe it’s my hearing but the shutter on the A7II is not that bad. Seems a lot more quiet than my 1Dx. Someone else was talking about shutter vibration but I did not think that was even an issue on on the A7. I still will be waiting for information on the A9 but if not for that I would probably venture into the E mount with this camera. There is a lot to like about this camera.
This what I couldn’t understand as well, to me the A7 II sounded twice quieter than the A7 and obviously much quieter than the A7R. I don’t see why the reviewer said it is the same.
Perhaps in this version the shutter noise goes primarily backwards towards the photographer and he heard the same loud shutter, while the recording camera mic recorded a much quieter sound.
The a7ii definitely sounded quieter to me. Add the improved ergonomics, AF and IBIS and it’s a solid upgrade as far as I’m concerned.
It should. The body is thicker now. Thicker body dampen sound better.
1dx has a flipping mirror.
“Would be nice to know from Sony why those lenses cannot get the full 5 axis stabilization…”
Answer: To make Sony more money. The answer to every questionable decision Sony makes.
If you were right (you’re not btw), how exactly would sony make more money with that?
Hopefully sell more of their own lenses and introduce full 5 axis on the A7 III.
God you are dense.
Trolls are really running out of material. Pathetic!
The goal of any business is to maximize profits, it’s called capitalism. Competition is the check on excessive profitizing, so you may have accidentally fallen on a weak point (even a broken clock is right twice a day). No other manufacture currently is offering direct competition to many of Sony’s photographic products.
If you read the comments you find that out, so let me repeat it for you. Manual and hybrid lenses give no feedback to the camera for the distance to the subject, only Sony native lenses and lenses made for the system by other manufacturers, also Minolta and Sony lenses that give a feedback by LA-EA Adapters will get 5 axis stabilization. All Leica and Zeiss M lenses or others used with adapters that do not communicate data to the camera will get 3 axis only. Ancient Sony bodies had 3 axis only and it was very efficient, so the benefit of 3 axis on all hybrid lenses is certainly a big step forward. It is certainly better than no stabilization at all.
That’s such a noob thing to say. Things boring over at Fuji and Oly camps?
This is all speculation. What Brian Matiash actually says is (and I quote): “The fact that it’s in body means that it’s able to be equipped with just about any lens you put on, to a degree. Depending on the third party lens, you might not get all five axis image stabilization, you might only get three…”
Which, if true, alludes to the fact that some third party lenses will, in fact, get in body 5 axis image stabilization. Now maybe he’s referring to the Tamron and Sigma A-Mount lenses via an LAEA-4 adapter, which, it would stand to reason, will function just as any other A-Mount lens would.
But more likely than not, he’s just speculating. This is a guy who’s had the camera for 1.5 hours, and he’s definitely not reading a script the way that a SONY manager would be when announcing a product. He’s a SONY artisan, sure, but he’s not their PR department. He may have just misspoke, or worse, he may just be commenting on what he’s heard on forums (like all the rest of us).
This is by no means a conclusive statement from SONY saying that all third party lenses will be limited to roll, X, and Y image stabilization.
Admins summaries are a bit weird sometimes, and then he even makes a headline out of it. I really don’t know why.
He is a Sony Artisan. I am 100% sur Sony epxlained him how the camera works 🙂
@admin
What the artisan says:
“Depending on the third party lens, you might not get all five axis image stabilization, you might only get three…”
What admin says:
“sony-artisan-brian-matiash-confirms-third-party-lenses-get-3-axis-stabilization-only”
So you talk about ALL third party lenses, including the native E-mount lenses from third party manufacturers like Zeiss (Loxia , Touit) and Sigma (AF lenses). No “might”, no “depends”, you mean all!
If these lenses get 5-axis stabilisation (what I think) then your headline is not just misleading, but simply wrong.
I’m usually not a nitpicker at all and I know that you are not a native English speaker, but these headlines on a site with many thousand viewers and being a source for many other reviewers you just have to be more careful with what you say. And if you quote other people, it better be correct!
Regards
Hans
Ok. To spcify I corrected the header.
I agree with Hans’ point.
This is a nice website and you’re a nice admin but lately you’ve been making some claims that are too out there… this affects your credibility. Even I now take your headlines with a pinch sodium. Under sell, over deliver… way better.
If you scroll to the very bottom you’ll see “powered by WordPress”. That’s simply not true. This website is powered by controversy 🙂
Corrected the title.
Sony
artisanassfanSAR_Admin is correct. The 5-axis IBIS works only if a lens reports the focusing distance back to the camera. Otherwise it’s 3 axis only. The new Canon lenses like 50 f/1.2 and 85 f/1.2 should work in the 5-axis mode no prob via Metabones. All Zeiss MF glass will work in the 3-axis mode.
No statement from Sony required, only common sense, 5-axis correction includes horizontal/vertical translation and thus requires the subject distance to known. Fully manual lenses have no way of feeding this information to the camera, hence you can only have 3-axis correction.
It still not bad. All Sony needs is to add another menu to allow users setting an approx distance to the subject, same as they did with the focal length choice, and all the five axis will work. Not sure if they’re gonna do it though 🙂 Seems like somebody needs to hack the firmware…
That would be tedious, think about it. If you have a moving subject the distance is constantly changing. My fingers aren’t that fast. LOL
Agreed, but most of the 3rd party lenses are MF anyway (even if they support AF – it’s crazy slow) and won’t work for the fast action anyway, unless you pre-focusing. I’m not sure if the distance option would be convenient, but it’s still better than nothing 🙂 Let’s say if the 3-axis IBIS gives 2 stop max benefit, on some occasions if you _really_ need the 5-axis’s 4.5 stops – the manual distance set up would be helpful.
It could be useful in certain situations, for instance at a concert where you would be taking shots or video from a predetermined spot where the distance from the stage wouldn’t vary by a great deal. It would be worth the extra effort to set the camera up prior to beginning the shoot.
Good point
A clever adapter could be made to report distance for rf coupled lenses… but meh who cares… 3 is plenty.
Good idea for a Kickstarter project 😉 Never is plenty.
I think this problem will resolve itself when higher quality native lenses are released. The bigger issues right now is the corner softness with RF lenses… honestly I don’t think such an adapter is worth it for the added axes that people never really needed anyway. Sony’s OSS also needs to be improved heaps… I get far better stabilisation from some of Canon’s cheap wide primes than I get from Sony’s expensive FE primes.
Translation correction makes a small part of the total correction and is only approximate (you can’t compensate for the apparent rotation of the subject as you move the camera) even when you have the exact distance, so it’s not worth the effort of constantly trying to estimate your subject distance. If you’re that bothered use a tripod.
Remember almost all lens based systems are 2-axis only.
Yes… and I have better success with optical stabilisation (adapted lenes) than what some of the A7II demo show with the full 5 axis.
Yes, only as in “only”.
Sony “artisan”…lol
@Admin
Is it possible to assign steady shot on/off function to a button?
Same thing is also stated by DPReview in their latest comparison, without any mention of Brian. Guess they got it independently.