Sony showed a prototype of a new CMOS sensor. Here are some of the specs:
- 17.7MP effective pixels
- 19.3MP full pixels
- 120 fps @ 12 bit or 60 fps @ 14 bit
- pixels pitch: 4.2 x 4.2μm
- size: 24.3 × 12.8mm (diagonal: 27.5mm)
- low power consumption
- reduced rolling shutter during video recording
- production date not yet disclosed
Related posts:
- Sony develops next-generation back-illuminated CMOS sensor with RGBW coding and HDR movie function, Canon files a patent for a large BSI sensor
- Sony Exmor R sensor now in digital cameras
- Yes, Sony Exmor R sensor is coming to Sony DSLRs
- Sony creates world’s first 16.41MP Exmor R back-illuminated CMOS sensor for mobile phones
- Next Sony DSLR: with the new Exmor R CMOS Sensor?
17 Comments
i guess another crop sensor but at least its good news for us Nikon shooters, maybe a new breed of cameras in the oven
but with how the situations in japan right now i guess everything will be put on hold until they get stable electrical supply and ground stop shaking for awhile though
Could This Be the Sensor for the D400??
Too small for the D400. This sensor is slightly smaller than DX, and it’s not 3:2 aspect ratio. It is for a video camera.
I don’t understand the 1.6MP going into places like “optical black area” thing.
Could someone explain ?
That extra area is there for calibration purposes, so that relative black can be determined. Also, the sensor requires extra area for ‘look around’, which is necessary to show on the evf for framing purposes.
I think you missed a few things in the announcement. This is a video sensor, not APS-C. It’s designed to produce the NHK Super Hi-Vision format, which is sort of a digital Imax (screens that produce 100 degree angle of view). As such, the sensor actually falls short on the vertical axis, which has to be line-doubled to reach the Super Hi-Vision standard (which is 4x 1080P in both directions).
This sensor won’t be used in still cameras. Some of the techniques used to get more bandwidth out of CMOS data exchange might make it into a new still sensor.
Thom, based on this comment (Google translated) they imply that this sensor technology can be expanded and used for digital cameras:
“As a first target, NHK Super Hi-Vision assuming promoted. Has about 16 times the information of the current full HD, it is known to be used for broadcasting in the near future. However, the technology itself that it can expand this flexibility without sticking to the size and number of pixels. Therefore, it can make a variety of specifications for digital cameras, ranging from mobile phones.”
In that comment they’re referring to the method of off-loading data, the bandwidth out thing I mentioned. Thing is, these days there are two core elements in a sensor, the sensel, and the support electronics. Once you have good designs for those, you can scale to other sensor sizes pretty easily. We’ve been watching Sony Semiconductor move multiple technologies across their sensors for awhile now, some that came from R&D outside the company (including Nikon).
So, could they put 120 fps data transfer into a future APS-C sensor? Sure. The problem will remain the same for the still cameras, though: since the desired output is 1080P, you likely end up line skipping and sub-sampling the horizontal axis, just as the current sensors do. All you’re getting is faster frame rates. 120 fps gives you somewhat better contrast AF, but not a lot more that’s really useful. Yes, I realize that you could do slow motion, but besides getting the data off the sensor you also have to process it through an external imaging ASIC and compress it. I’m not betting that we’re near any great breakthroughs that’ll give you overly compelling quality at 120 fps with a consumer DSLR.
@ Thom
The sensor can do 8K4K using vertical interpolation (debayering on y-axis), which is SIXTEEN times 1080p; 7,680 x 4,320 pixel resolution Quad-1080p is merely 4K2K.
This camera doesn’t do debayering in the traditional sense, like the F64 cinema camera Sony announced, instead it combines RGB pixels into one, which doesn’t have the effect of inflating resolution numbers, color information is combined at the photosite level not after the ADC. Which is why the 8K sensor produces 4K raw files. However, these ‘Q67″ sensor can also do vertical interpolation to make a 8k4K image in a similar sense to the traditional bayer. Sony called this a “double Bayer’ at NAB.
Let’s keep in mind, this sensor has less performance than the F65 sensor that was just shown. It has less pixel density, and the F65 had a full 16-bit ADC, this sensor has a mere 14-bit ADC.
As far as this sensor’s implication to dSLRs. This new duel-row readout ADCs will likely find its way into dSLRs. We may even see these new CFAs in dSLRs as well.
Perhaps I should have been more clear: by 4x 1080P on both axes I meant 4x 1920 and 4x 1080, which is 7680×4320. Sorry if this was misread.
My understanding is that Sony has turned the underlying Bayer pattern 45 degrees, much like Fujifilm, but I only heard that second hand. That’s an interesting turn if it is true, as it puts the maximal resolution on a different axis than Bayer (horizontal not diagonal).
I agree that the dual-row readout will make it to DSLRs. Not sure about any of the other stuff, though.
This new sensor is not similar to the bayer CFA rotated 45 degrees (Clearvid in Sony-speak). The first indication should be that this sensor doesn’t do the debayering or interpolation (which results in 8K4K in this sensor).
http://image-sensors-world.blogspot.com/2010/12/sony-cinealta-moves-beyond-bayer.html
This new ADC allows for the combination of parts of the R, G, and G photosite to form panchromatic pixels. This is very different from traditional bayer sensors that convert an analog signal into either red, greenx2, blue pixels off the sensor that then must be debayered into a usable image. This has an effect of inflating the resolution (more megapixels) of the image, but is using a third of the color resolution to form what is essentially less than half.
These new Sony sensors are combining photosites at the ANALOG state to form panchromatic pixels rather than pixels that are monocolor. This is happening before the ADC touches it.
This is only possible because this new ADC is able to sample multiple pixels sententiously rather than row-by-row like a traditional rolling shutter CMOS.
This sensor really is moving BEYOND bayer…
pixels pitch: 4.2 x 4.2μm
crap
?
Similar to a Canon 7D in terms of pixel pitch…
@mog the 7D is not a good camera, cosmetic is good and selling but not IQ
7D is a fine camera, dSLRs in the ~4 micron range are fairly common, the image quality is fine. Besides, as technology improves pixel pitch has become smaller as low-light sensitivity and DR has improved.
If massive pixel size is what you’re after, Sony has the F3 which has 12-micron+ pixel pitch.
you like the 7D ? very well
did you compare the files with something else ? I mean not with a compact
4 micron is common for familly-like photography only
The Nikon d7000 is in the 4 micron range as well microns in pixel pitch, and isn’t far off from the 7D, it’s like o.4 microns away in terms of size. It’s common in dSLRs.
Also, this sensor bins pixels, 8k is binned to a 4k(around 8mp). So it’s acting like a 6 micron photo site.