Sony sensor tech. Image source: VLSI Symposia tipsheet
A couple of weeks ago I showed you two patents describing a Full Frame curved sensor and a 35mm f/1.8 lens designed for such a sensor. Usually when you see Sony patents you well know that it could take ages to see it applied on a real world mass product. Well this could be one of the few cases where a patent becomes “real” in a very short time.
First of all the prestigious site Nikkei Japan reports that Sony has developed a curved sensor that doubles the sensor sensitivity. And Nikkei doesn’t spread a news based on patent 🙂 It means Sony is really ready to launch that kind of tech!
And than VLSI Symposia tipsheet (via ISW)explained better how this sensor works:
“Curved CMOS Image System: When light transmitted by a lens strikes a perpendicular target such as a CMOS image sensor, it forms a circle of light called an image circle. It’s difficult for a flat (planar) CMOS image sensor to deliver high image sensitivity at high resolution (highly scaled pixel pitch) because of the fundamental physical limit known as quantum efficiency. To break through that physical limit and to achieve higher sensitivity anywhere within the image circle at higher resolution, Sony built and will describe an imaging system that comprises a hemispherically curved, back-illuminated CMOS image sensor (BIS) and integrated lens. It doubles the sensitivity at the edge of the image circle while increasing sensitivity at its center by a factor of 1.4, with a 5x reduction of dark current (Jd) compared to a planar BIS. Moreover, a common problem known as lens field curvature aberration (Afc) is mitigated by the curved sensor itself, and so the curved BIS enables higher system sensitivity with a brighter lens with a smaller F number (Fn) than is possible with a planar BIS. In addition, by controlling the tensile stress of the BIS chip to produce a curved shape in the first place, the energy band-gap (Eg) is widened and a lower Jd is achieved. (Paper T2.1, “A Novel Curved CMOS Image Sensor Integrated with Imaging System,” K. Itonaga et al., Sony)“
And as you know an anonymous source said the RX2 with curved sensor tech and 35mm f/1.8 lens will be announced in summer. it has to be seen if Sony can use such a curved sensor on an interchangeable E-mount camera…
0 Responses
Probably they need it to enable fast wide angle lenses for FE mount.
to achieve what? planar objects eg. buildings will still require small apertures, it’s probably there to eliminate the distortion due to perspective of WA lenses.
To get rid of colorshift and pixel vignetting.
Colour shift is due to lens design. (Telecentric lenses don’t cause it). Vignetting is related to focal length for example Zeiss 15mm f/2.8 has rays hitting the sensor almost directly, it has not colour shift but vignettes due to short focal length.
Colorshift occurs in the sensor, not in the lens. Sure, it can be avoided with image space telecentric lens, but telecentic lenses lose in IQ/size terms.
This feature would likely be available for all interchangeable lens systems. It’d probably be installed on Sony’s best sellers like the A5000, A6000, or A77III well before a niche products like the FE mounts.
I’d be very surprised if this ended up in any ILC systems. It’s perfectly suited towards a fixed lens camera though, and my guess is that that’s where it will end up. Using this on an interchangeable system would require completely new lenses, which Sony obviously isn’t the best at. Granted, the lenses would be much easier and less costly to make as you wouldn’t have to correct the image the way you do with current lenses, but still…
Will past lenses be compatible? or does it require new design?
Thats the goal of curved sensor, having new lenses. More light, less glass, better and so on.
Having planar films or sensors was in the last 150 years only a matter of the absent tecnologie!
Fixed lens.
This seems to be a good idea, but can it be implemented in every camera ? and do we need new lenses again ??
Ooo maaannn… I most of the time can’t resist curved hips, now sony tempted me with curved sensor… Be stroonggg be strooongg…
That’ll mean accurate AF points will cover most/all the frame rather than being clustered around the middle?
This design doesn’t provide mirror AF at all, so AF at all sensor will be accurate xD
“curved BSI sensor with integrated lens” – it looks like for the smartphones/tablets only.
Sounds very interesting. Although it makes me predict that Sony will introduce a new FEC-mount: Similar to FE mount, but with curved sensor. FE lenses will (of course) not be compatible and Sony will stop producing new FE lenses (as it did with A-mount lenses and E-mount lenses before) but focus all its attention to the new generation of FEC-lenses.
Seriously, I still think the tech is better suited for small chips like in smartphone cameras, but it would certainly give a terrific RX2. Considering that I barely ever take off the 35mm from my NEX7, a fixed focal length camera sounds more and more interesting.
Although rx2 is likely to be ff, so it’d need a 50mm instead… If was 50mm ff I’d be very interested, 35mm less so (I use 50mm equivalent more than 35mm equivalent on my a77).
why would sony make an expensive ff sensor just for the use on rx2? i think the rumors about rx2 is not correct. The curve sensor will likely be a small sensor for smartphone
The patent specifies a curved ff sensor and a 35mm f1.8 lens.
As far as I understand the principle, you will have to match a lens and a sensor to get really the best possible result, like it may be the case in a future RX 2. Also a zoom lens could cause problems. Therefore, I doubt that there will be any interchangable lens camera equipped with a curved sensor; unless Sony develop a “variable curvature” which seems to be impossible given the brittleness of silicon. Its more a tech for fixed focal length cameras, no matter if this will be a high end like RX or a smartphone.
I don’t think variable curvature is what they want to make but having worked in the Silicon industry for many years I can definitely say that Silicon, if thin enough is flexible.
A wafer can bend more than you imagine. Furthermore Silicon are even used as springs in certain applications.
A small company in Germany(Holm Silicon) has for instance specialiced in making Si springs for watches among others Omega watches.
But with all that being said a sensor is not only Silicon so I think it is impossible to make a variable curvature sensor.
Providing a variable curve that remains acurate under working conditions would introduce reliability issues.
I have speculated about what curved sensors could do as well as tilt and shift capabilities. If we are to dream…let’s dream big. How about this in the future? Integrate sensors in the lens so you have the optimum designed sensor per lens. Of course we would also need a global electronic shutter and manufacturing costs of the sensors would have to come down, but we can dream.
Does anyone here understand the “Field Curvature” aspect of this post? Is it related to the field curvature that we see from lenses like the Zeiss 24mm 1.8 E where the focus field travels towards the distance as it leaves the centre of the frame?
Would this be the same FC? or a different one in relationship only for the lens to focus on the sensor itself that does not represent the image characteristic I described above?
Nevermind. Having a second thought about this it would appear to be that shit optimization is non-related to lens drawing characteristics but rather about making the abherrations seen on sensors today disappear as if the days of film where only the lens had the final drawing characteristic of the photograph an not as today where a lens can perform differently on different cameras. My bad.
We are dealing with a 4 element apochromatic lens desighn.
When this is symetrical…back pair being identical to front pair with aperture in the middle…the lens will produce an apochromatic flat field 1/1 image…but at other magnifications the quality decreases.
By making the lens asymmetric it now performs better at the magnifications commonly used but the trade off is that the focal plane is no longer perfectly flat.
By curving the sensor to match tis curve the superb image quality is restored.
Also..the light angle of incidence across the sensor approaches 90% and the distance to the lens becomes equal therefore eliminating light drop off.
So do we need special lenses for this sensor? How would the current designs perform?
In order to maintain a flt focal plain at we would need lenses with fields that match the sensor curve.
Todays lenses are corrected to flatten the field.
This does not however mean that these lenses cannot be used..they may for example produce an image where the center of the subject is in focus but the edges focus furyher awy…or just the opposite where the center is infocus but the edges focus closer…usable f poo r special effects but not for producing accurate images.
In short…this system is great but presents setios problems for interchangeable lens systm.
It will not appearon ilc cameras soon if at all.
There is hope then for RX 35mm, 50mm, 24mm, 21mm, 85mm with amazing optical qualities and, as we already know, great sensors. This is the kind of tech that can get people to buy a Sony “series of cams”
Yes
Any focal length or magnification can work with this technogy but interchangeability would be an issue.
Interchangeability would only be an issue in the near term. Eventually sensors will be made to curve on demand to meet the need of the lens or even as the focus distance changes.
A variable curve sensor would need to reliably and preciselly alter its curve on demand for the lifetime of the canera.
Its too much to ask a sensor to do even if the technolgy to do it were there.
What I will predict for this camera..strictly speculation..
Resolution will exceed 100mp
Firmware will include perspective control.
This system will work better with bayer array than stacked.
Curved sensor is likely to have barrel or pincusion distortion.
Bayer requires interpolation..demosaicing…therefore perspective correction will not degrade the image…since the firmware requires that ability it could also be available as a user controled tool.
It means that ILC’es are 2nd priority for this technology.
This system would work best in P&S and mobile system applications. It’s not really a system pointed towards ILC designs.
Barry please stop sticking the word ‘apochromatic’ everywhere.
What the f**k is your issue.
Apochromatic is exactly what we are talking about.
Both the front and rear pairs are achromatic sets…mathed for 2 colors…by reversing them the 3rd color is alighned…look up apochromatic…if you cant handle the word get out of photography.
Also look up symetrical apochromatic lens you might actually learn something.
You are the only one who decided curved sensors have something to do with apochromatic lenses. They are not.
Look at the patent drawin.
It shows a 4 element apochromatic lens..if you read what I asked you to read you would recognize the desighn.
This is the same Field of Curvature. The focus distance of a simple lens depends on the distance of the sensor and the lens focal point. Since it varies between center and corners for a flat sensor, lenses a) need complex correction and b) still have FoC issues. Making a curved sensor thus simplifies lens design greatly. Complicates sensor design though..
Back illuminated sensors have a smaller focal length to the microlense.
In a lens the f # is a ratio of the focal lengh an aperture.
Therefore bis sensors are effective in less light..meaning …smaller pixels and higher resolutions are not only possible but more practical with them.
Bis sensors are traditionall used in small p&s cameras where the pixel density is much higher than on larger conventional sensors.
If this ff bis sensor is built with the same pixel density we are talking about 100-200mp.
The lens being a simple 4 element assymetric variation of the old symetrical apochromat will be one of the sharpest lenses ever made and can effortlessly handle the high resolution this sensor desighn is capable of.
I do not know what resolution sony plans to put into this camera but if they want to this could easily over 100mp.with superb quality and good low light performance.
And on the day it’s released, we’ll still only get 24MP, just to upset you Barry. I think the Sony guys have it out for you.
You coul be rite but back illuminated sensors are specicall desighned to perform well at higher prxel densities and do not perform any better at low densities.
Therfore 24 mp would be a total waste of this technology.
Nah, I’m just joking around with you.
I’ve been dreaming of a curved sensor for very different reasons than image quality: autofocus accuracy. Be it contrast or phase detection, the autofocus has usually it’s best low light ability in the center. Now, if you pan the camera with a curved sensor after focusing, your focous stays in place since the plane which is in focus also is curved. You agree?
I think you are wrong.
Most probably even with these curved configurations the designers will make sure that the plane of focus is as plane as possible as that’s what most people will expect. For instance, shooting landscape would become quite cumbersome with such an output.
What you are looking for are lenses with field curvature and most lenses today will show it to some degree. In general it is considered an optical defect. There are some lenses like the minolta VFC (you may want to google it, its very interesting), that allow you to set the amount and type (convex/concave) of the curvature. Since Sony should basically have access to the patents in principle they don’t need anything additional to create your dream lens.
No, the focal plane is still going to be flat. And, it will give the same parallax errors with focus and recompose. this just makes it easier to design lenses
This illustration is intersting in that it depicts a traditional camera reasonably accurate but depicts tthe curved sensor camera as a sphere with a single element lend to showits similarityto the human eye.
This will actually have a 4 element lens..not a 1 element lens.
this tech reminds me how human eye works…curved
I totally agree.
The visual resemblence of this drawing to an eye is to illustrate the functional similarity.
I will order a 7S with curved sensor (or without it), but with 4k recording in the body.
So the 7s II…in 3 years.
Yeah. And not be able to successfully use any legacy or third party lenses.
iphone 7s with curved sensor! 🙂
…this kind of sensor will change the camera market, I think. Maybe street photographers will take their images with mobiles…
Will we need curved monitors and curved prints?
No and no. No more than you need them now.
“it has to be seen if Sony can use such a curved sensor on an interchangeable E-mount camera…”
Completely new lens system will be needed to take full advantage of curved sensors. Meaning, Sony will introduce yet another lens system without finishing any of the existing ones. 🙂
Unless it is a fixed lens camera, then they have to design only a single lens dedicated for this camera, as per the RX1, RX10 and RX100, as the rumour also suggests mentioning a RX2 camera! Also it is supposed to be a simpler lens having to produce a curved image with less corrections required.
You’re thinking camera systems, but you should really be thinking point and shoots (a la RX series) and mobile systems. That’s the most likely implementation for this type of sensor, as the sensor would work best when matched to the lens.
Again, see what I am answering on – “it has to be seen if Sony can use such a curved sensor on an interchangeable E-mount camera…”
This would be a really big breakthrough, mainly with the lenses needing less corrections and respectively less elements, resulting less glass to air surfaces, less light loss, increased contrast and resolution and compactness and reduced price!
Definitely! I’m quite excited just reading the benefits/possibilities of this curved sensor….you gotto hand it to Sony for innovation (compared to C&N)
minipixels.net | Automotive Photography + Design
Reduced cost for the manufacturer maybe, but hardly reduced price for the customers.
Since it will be competing against many established camera systems the price can not be unreasonably high.
Well, let’s hope you are right.
How about Ricoh GXR type of interchangable lens+sensor? If the total costs of producing such componet is comparable to lens only, then it makes sense to have 2 camera bodies ( one for Video, one for still) with only viewfinder and controls, sensor/lens sizes then depend on the application, e.g. HD video lens can have a 28-300mm T/2.8 lens on a 1″ sensor, 4K video 24-120mm T/2.8 on APS-C sensor. 35/1.8, 85/1.8, 200/2.8 on 35mm sensor, 28/4, 55/2, 120/4 Macro on a 44x33mm sensor… And perhaps a 150-600/5.6 superzoom on a 1.5″ sensor!
This is simple!
They work on a 1` 4k sensor for Samrtphone and a small fixed f 2,8 Lense.***
Because Sony survives the smartphone-4k-cam-hystria ( all its inhouse-camlabs
wil be closed)- ore short :: you alwas work for the market where you earn $$$$$$$$$$$$$
and more $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
***
artifical small formfactor and yes with the design-4-corners and 120 meters water resistant
and…. if this goes for 19,90,-usd$ / month…would you buy a camera?
……..nope? okay…. $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
Dude, try to organize your thoughts before writing them.
It’s name is “patrik not jane”.
There’s more on it’s List than organizing it’s thoughts. Spelling being one of them, thoughts about it’s organs being another.
I agree that this implementation is almost tailor-made for phones.
I still wonder what the bokeh will be like from such a system.
Can this technology really be used on an interchangeable lens format?
Every lens has a different field curvature, and the light hits the corners at a different angle for every different lens.
I believe this is only practical for a fixed prime lens system like the RX2, and we will never see this in an interchangeable lens camera.
Would this cause problems for focal plane shutters at very high shutter speeds? Unless the shutter was also curved, different parts of the sensor would theoretically get slightly different exposure times. It’s not like we have central leaf shutters in anything with interchangeable lenses (except Leica S).
Impossible for A-mount nor E-mount since we know that there will never a back-illuminated CMOS planned for these mounts (Sony said).
Still one fantasmatic rumor based on a personal dream?
And one hell of a technology to shrink down integrated lenses (RX1, RX10, RX100).
Well you can’t shrink anything for the A-mount and no more for the E-mount…! RX2 is not another fantasmatic rumor? 🙂