Sony patented a new sensor having two layers. It has one classic RGB pixel layer on top and one phase detection pixel layer below. This means that the phase detection works on every single R or G or B pixel and covers the full sensor area. Quite amazing! Let’s hope this one will be implemented on a real camera n the near future.
via Egami.
80 Responses
I see the pattern:
1. Remove the conducting material to the back of the sensor (BCI) to allow more light to hit the sensor
2. Make quantum film to allow more light to hit the sensor
3. Slap another layer on top of the sensor
No, the Phase Detect Layer is below the RGB Layer… So, its not impacting the Photo Sensitive Imaging Diodes…
“It has one classic RGB pixel layer on top and one phase detection pixel layer below.”
There are two pd layers and a white pixel layer.
Abmin is mistaken in his description of this patent.
This would be an excellen choice to utilize the quantum thin film technology on this sensor, an I hope they do.
With miultiple layers the quantum advantage would be even greater than on a concentional bsi sensor.
As far as I can tell thogh the patent does not specifically indicate QTF…quantum thin film….will be used.
Is it limited to a mount somehow and impossible with e mount >:I I hope so, a mount foreva!
XD
Whatever the particular caveats, surely this sensor addresses both AF speed and accuracy surpassing that of all other forms, at least considered combined together.
Unless someone can point out how it is not the type of process used with mirror and such, nor ever potentially as effective…
Its exactly what I said as a layperson not knowing much regards how AF works, and on the matter of no mirror or deflection while at once gaining per pixel accuracy. Again, isn’t that what has started with the new bsi sensor with many phase detect points?
Your A mount camera line no longer has anything but materialistic meaning, like the Hasselblad converted Sony’s, buy them up now before they become paradoxically collectible!
:]
With suitable and effective algorithms, this may allow AF in shooting video at high res too surely!
Master photographer. 😀
Another nail in SLT and SLR coffin? At some point OSPDAF will catch up with dedicated PDAF sensor speed and faster sensor read-out speed will eliminate EVF lag and black-out. When that happens, SLT and SLR will make absolutely no sense. Call me skeptic, but I no longer expect any advances in A-mount.
I think the next A Mount may already have the mirror removed, and be strictly PDAF. Maybe that is why the A99 replacement is taking so long. Waiting on the sensor tech to mature. We should know by about September, one way or the other.
What real purpose, apart from satisfying remaining loyalists, would such A-mount body serve? FE 90/2.8 macro, FE 35/1.4 lenses and A7rII OSPDAF are clear messages about where Sony is heading.
Why create A99 with removed mirror if they can do similarly sized (ergonomics, battery, controls) A9, which would sell better.
That is what I was talking about. A9 will most likely be A Mount. It does not have to be as big as, or look like the A99 to be the replacement. It is also possible that it could be universal, with 2 different mount modules, one for FE & one for A. This would also make sense for people who have lenses from both systems. Remember the rumour a few months ago that someone saw a prototype of a universal mount camera from Sony.
Even if they do another generation of A-mount cameras (less and less likely), it will be the last. Once the AF and sensor read-out speed is sufficient, A-mount has no advantage over E-mount, while E-mount will still have advantage of adaptable lenses, possibility of smaller bodies. They will not keep developing two systems just for some remaining loyalists. Don’t get me wrong, I am still using A-mount with Zeiss/SonyG lenses and love it, but I have A7rII, Batis 85, 90 macro, 55/1.8 and LA-EA3 for my SSM A-mount lenses on order. A-mount will not die, but the A-mount cameras will. Accept, move on 🙂
I am in the same spot. I am told that my A7RII will be in the shop on August 5th, and my LAEA3 is on the way from Amazon in the next few days. Once I see that the system works well with A Mount lenses, then the A99 is going on Craigslist/Ebay right away. The only thing missing for me is an LAEA6 that will work the screw drive lenses without the mirror. I have the LAEA4 already, but the AF is kind of a poor implementation. I hope Sony steps up with a better adapter. I like my 85 1.4 & Minolta 200 2.8 and would prefer to keep them for a few more years.
I would love the adapter without SLT but with focus motor, and they should have released that along the A7rII, but only non-SSM lens I have worth keeping is the 135/1.8.
In the end, I am quite excited about the switch – it will allow me to have single system, with best possible IQ, for both light travel and professional work – something none of the competitors can really provide.
Sony wont come out with an Adaptor for you to use your amazing screw drive Minolta lenses. Why do you think they dont have one already?This was the whole reason Sony switched mounts. SO YOU WOULD HAVE TO BUY $1,000+ toy lenses.
“Even if they do another generation of A-mount cameras (less and less likely), it will be the last. Once the AF and sensor read-out speed is sufficient”
Yes, once AF is as good or better than there will be little reason unless it is still profitable to produce one or two a-mount bodies. For now, there is sufficient demand for an a99-2, particularly if it can be made to be their fastest camera. The e-mount system still has a ways to go.
I agree. E mount is a school age kid. Still need to mature more to be an adult
It’s not just that they’re waiting for AF to be just as good. We’re talking tens of thousands of users already invested heavily in A-mount. As long as they’re going to keep buying, what sense would it make for Sony to stop producing?
Agreed
The A9 Will be A-Mount. Sony will not have an “A99ii” It will be the A9. It will be a new start for A-Mount in FF like the A900 was. Its even a play on the A900 modeling naming, Just minus the 00.
Releasing updated a-mount lenses strongly suggests that Sony still thinks there’s life in the system. The problem with an a9 is that you would still need to adapt lenses that don’t natively exist in the e-mount lineup. F1.4 suggests a certain direction of travel, but we’ve not had a hint of e-mount f2.8 zooms or longer tele-primes yet… Personally I’m more than happy to adapt but I’m an unashamed hobbyist. A pro, on the other hand, might see that as too big an ask. The option to adapt your existing lenses is great, but requiring people to do so to plug gaps in the system isn’t.
Sony said themselves they realise people want f2.8 zooms.
But have been suspiciously shore to do anything about it. My guess is as soon as they do, they know it’s time to switch off the a-mount’s life support…
They said it what, a whole month ago? System has been out for what? 18 months? Oh and they just released a camera that works well with the brand newly minted updated versions of their two existing f2.8 ZA’s. Yes the situation certainly seem dire.
First admitted it publicly a month ago. If they’re talking about it in public, you can bet they’ve been discussing it behind closed doors for much longer.
What did Sony “admit?” I must have missed it . . .
They talk about it here: http://www.dpreview.com/articles/0717419525/interview-kimio-maki-of-sony-the-customer-s-voice-is-the-most-important-data-for-me
Sorry Andrew. I must be thick. I read that article before, and I just read it again. I don’t see anything which I would call an “admission.” What do you think Mr. Maki admitted?
so our next lenses will be at the upper end of the lineup.
In terms of focal length?
KM: Exactly. And also aperture. A F2.8 lineup is necessary
OK. Thanks for the clarification.
And people says I’m insane when I suggest A mount with full OS PDAF and no SLT mirror a few months ago in SAR forum. Anything can happen with Sony~
I think after a few years, the only difference between A & E Mount systems will be the Body Style. Everything will be an E Mount camera with OS PDAF.
The DSLR styled body will come with an LAEA3 Style Mount adapter maybe with the AF Motor for Minolta Lenses.
If it’s an E-mount “DSLR styled body” it would be bulkier than a DSLR when you add the adapter. No thanks!
This will have substantially faster and more accurate autofocus than any slr or slt.
If you’re so wanting to see end of DSLR, you might imagine as well sensor which can act as matte glass in a DSLR
You are a skeptic. I expect a-mount to continue and thrive for the foreseeable future. Any relevant advances and innovations will be carried over to a-mount. A-mount will be Sony’s top end line and e-mount will continue as a second tier line which will sell in larger numbers at cheaper prices.
um to date, the most expensive cameras are e-mount, not a-mount (a7rm2 and videocams)
Camera bodies, yes, at least until the a99 replacement launches. (And I am really concerned about what the a7rII’s sales numbers are going to look like at $3200 a pop.) Lenses, not even close.
there will be even more costly a9….
Will there be? And where are the native lenses for such a pricey beast (let alone the customer base)?
some kind of pixel binning at the phasedetection layer @ low light?
Put it in a A99II and with mirror lock up option for full OS PDAF glory~
(I don’t like the pellicle mirror and I don’t shoot sports anyways. My models don’t run around crazily)
And some says A mount glasses doesn’t work well with OS PDAF, but why no one says Metabone adapters is bs which I believe Canon lenses are clearly not designed to work with OS PDAF in mind. Please educate me, my ears are all in~
People have never really said that, just they aren’t good with CDAF.
No, they don’t run around crazily, they pretend to be lesbians according to your direction instead. I’m surprised that with all the bs I have to contend with, i’m yet still the only one who says it, and in future all the extra bs simply darts up and turns like battling wasps in my direction. You should get the nilos to work for you, they are weak too.
This sounds exactly the Canon technology found first in the 70D. The dual pixel auto focus is by far the smoothest and most organic-looking auto focus pulling in the market. I hope we see Sony’s version of this technology next year in a A9 E-mount camera.
42 Million AF Points… Holy %@#$!!
Just imagine trying to select the right one.
“OK guys I’m switching focus point. See you all after lunch”
I imagine they’d just have much larger phase detect points instead so OS PDAF worked in extreme low light while still covering the entire sensor. That would likely put another nail in the DSLR coffin.
Obviously, but that’s not very funny, is it.
I’m just gonna put my focus poit on that strand of hair over to the left… or maybe the one next to it…
you need more than one pixel to make a focus point (which isn’t really a point but a spot)
Just because a company takes out a patent doesn’t mean it will run into a real product. Very often these are just speculative ideas just in case they may turn out to be useful. It’s not even necessary to have a prototype to take out a patent. This particular idea raises a lot of questions, so I don’t think people should just assume it will turn into a real product. My suspicion is that it’s not very likely, especially as great strides are already being made with hybrid on-sensor AF systems.
The question is for what else can the phase information be used when it is stored for each pixel?
Maybe we can set a polarisation filter via postprocessing?
Can we calculate 3D pictures from the phase information?
Any other ideas?
Maybe we can even build a phaser?
Probably yes, you’ll be able to get depth data for sure. Maybe work with selective bokeh after the shot.
Interesting but not really a necessity for now.
RX100mk5!!!! Sweet.
I believe Sony is and will be in the sensor field for a long time bringing new tech often. The only thing I find odd in the design of E mount cameras is the lack of outputs. For example, I can’t wireless remote trigger the A7 while tethered to my laptop because there is only one Micro HDMI and is used to tether with the Remote Camera Control. Or the lack of wired flash sync port, I will need to use an ad-on on the hot shoe to make something work like that.
Those are simple features I find on my A77II and many other pro models. If Sony want to Canon and Nikon users to switch, they should provide a minimum of things they are used to work with without having to re invent the wheel. Size, speed and quality found on the A7RII won’t be enough for them to switch.
That’s why Sony still has A-mount.
You can still use the infrared remote trigger.
Infrared needs direct direct visual of the IR receiver of the camera
You could use WiFi to trigger the camera
Can you further explain what you mean by WiFi?
You said that you can’t trigger the camera remotely since there aren’t two HDMI ports. But the camera does have WiFi so you can use a smartphone or tablet to trigger the camera using the Remote trigger app already built into the camera.
I do understand the you can use the app to trigger over wifi but it won’t allow to tether and use the app at the same time. Sony needs to realize that and add this functionality so people who want to switch can do so without complains. I do have an A7 and that’s one of the problem that held me back to enter E mount. On the other hand the A77II do have these features
Ah OK, I don’t shoot the A7 bodies tethered so I did not realize that limitation. Does that limitation also exist if you use Capture One as the software you are tethering with?
I haven’t tried with Capture One but the problem resides in both apps not being able to work together in this case.
A-Mount is for high end uses. E-Mount for people who like it small.
Admin
I believe your description of this sensor may be incorrect..
As far as I know this is a 3d mulltilayer stacked bsi pd sensor.
4 layers are
Conventional bayer …color layer
1st phase detect layer
White pixel …..bw sensor layer
2nd phase detect layer.
The 2 layers of pd sensors allow vertical seperation therefore autofocus speed will be far beyond anything currently available.
The white pixel layer will increse sensitivity and dynamic range when combined with the coler image through special processing algorithms.
White pixels are automatically at least 4x as sensitive as color pixels because they are not limited by color filters.
The stacked nature of this sensor…processing elements in the sensor…will make this sensor very fast.
Thanks for explanation. Would you mind answering some questions :
(1)What does “…allow vertical seperation…” mean? Can’t the present OSPDAF do vertical seperation ?
(2)Why does this structure need a white pixels layer ?
Thank you very much.
Thanks for explanation.
What does “The 2 layers of pd sensors allow vertical seperation…” mean? Can’t the present OSPDAF do vertical seperation?
Besides, why does this structure need a white pixels layer?
Thanks a lot.
It is relatively easy to tell if an image is in or out of focus.
It is more difficult to know if the lens needs to move in or out to achieve proper focus.
Two layers of pd pixels are at the focal plane and behind the focal plane.
By comparing the data from the two layers the processor instantly knows wich direction and by how much to move the lens…resulting in much faster and more accurate autofocus than we have ever seen before.
That is the advantage to the 3d pd pixel array.
The white layer of pixels lies between the two pd layers.
This layer is far more sensitive to light than the color pixels bit lacks color information.
By combining..interpolating ….data from these two layers the image will be higher iso, more dynamic range than is possible from the color pixels alon…so the white layer determins the light intensity and the color layer determines the hue.
Because the sensor is stacked processing elements are included within the sensor that speed processing and allow all the extra data this sensor is accumulating to be processed quickly.
I do not know if this will be combined with the new quantum thin film..QTF…technology, but clearly this would be highly advantageous, possibly even necessary to make this sensor work with 4 layers.
the dual pixel share the same hard ware with the imaging pixel, which will reduce image quality because of doubled thermal noise, but this one from sony is fully independent. which raise up the cost for sure but likely no harm to the image quality
This is nothing like the Canon sensor – not even remotely similar.
A7V confirmed!
It seem to be a great tech, but they need to get fast AF beyond -4ev, that would be the real groundbreaking achievement.
A7000 with universal mount…..
This won’t work.
Why not?
From the looks of the patent, it seems to be a dual layer AF(admin is wrong, there are two AF layers, not one. 11a is the today’s layer, 11b is the additional PD layer). Top layer for judging the focus accuracy, bottom layer to work in conjunction with top layer, to know the direction in which the lens should focus and how much.
This sounds like they are trying to achieve the speed of DSLR with accuracy of mirrorless. Looks like a nice idea. Finally a mirrorless camera wouldn’t have to hunt constantly back and forth to figure out where to focus.
Thanks for explanation. The present hybrid on sensor AF is : PDAF for speed, CDAF for final accurancy. How can this 2 layers PDAF take advantage in both speed and accurancy over hybrid AF ?
The following is my understanding:
The CDAF can still be used if needed, but what you’re gaining in aforementioned PDAF config is the AF speed of a DSLR, without the typical mirrorless back-n-forth hunting, if this is implemented right. In good light, the PDAF that is embedded deep in the sensor, will receive enough light to be able to tell just how much out of focus the lens is, based on the comparison to the primary array and the f-stop of the lens. This can tell the camera body, how much the lens has to turn. This would have a huge impact on tracking and overall focusing speed capabilities. Exactly the advantage that today’s DSLRs have.
+/- 0.5mm at the sensor translates into meters between the lens and the subject. So a little distance between the layers is enough to tell how much out of focus the lens is.
Unfortunately this doesn’t improve the low-light AF capability, unless Sony decides to use some large PDAF pixels.
As Steve said, this is speculative patent. Light adsorption/scattering in upper layer will render bottom layers with very few light and it’s blue (shorter wavelengths) which gets absorbed by silicon better (in thinner layers)
They do have one already, it is called the LA-EA4, but it has a translucent mirror in place, and provides Phase Detect AF, but now with the new sensors having PDAF on them, we could do without the mirror.