(SR4) Compact cameras with “Pixel Binning” coming in early summer.
One of my very trusted sources told me that the next Sony compact cameras will have a new “Pixel Binning” feature. For those of you not knowing what this is here is the wikipedia description: “binning is the procedure of combining a cluster of pixels into a single pixel. As such, in 2×2 binning, an array of 4 pixels becomes a single larger pixel,[1] reducing the overall number of pixels. This aggregation reduces the impact of read noise on the processed image at the cost of a lower resolution.“.
This is an interesting news because it is also a feature many of you requested for the future NEX and SLT cameras. I don’t know yet if that feature will be exclusive for compact camera or if it will be implemented in future system cameras too. I would certainly welcome that feature on my future A99 
Related Posts
Share
Recent comments
Comment from mark: ...and it looks like Nokia will also use...
Comment from saturazzi: Wouldn't binning of four colour pixels (...
Comment from Street Wizard: Read about pixel binning here: http://ww...
Comment from Tom: Or... if you're feeling particularly dar...
Vivek
4 months ago |Why make tiny pixels and then combine them (doing a pixel binning) later to reduce noise instead making larger pixels in the first place?
shamb
4 months ago |Because you want high resolution in good light, and big sensor surface per pixel in low light.
Hello World
4 months ago |+1
passer-by
4 months ago |+1
monkeyfacemcbride
4 months ago |Versatility, in good light you dont bin and get the benefit of all thouse pixels, in bad light you lose the pixel count and gain the low noise. All this occurs on the sensor before the digitization phase. Its been used in Astro photography for a long time. Essentially you sacrifice detail for a reduced noise image.
You cant achieve the same thing in post as the Summing Well on the sensor has the benefit of a single readout channel. If you do it in post from the full pixel count the 4 pixels have 4 readout channels hence 4x the readout noise (which would add in quadrature so ends up being about twice the noise) so the results are still more noise and worse colour accuracy. True binning is better if you can stand the resolution drop.
Bugger
4 months ago |Versatility..
PhotoNut
4 months ago |Who cares about compact cameras with tiny sensors? I made the mistake of buying the first Sony compact camera with backside illumination sensor and CZ f/2.5 (or something) lens thinking it could tidy me over in cases where I didn’t want to carry my bigger DSLR. Big mistake. the POS didn’t even have any way to shot aperture or shutter priority. Everything was “auto” mode. I was so p*ssed off and swore off buying any Sony compacts again.
Till today, Sony does not make any compact cameras with bigger sensors or any compact cameras with support for RAW.
These new compacts may have pixel binning but I can guarantee you that they will never output RAW images or let you control the level of noise reduction applied meaning that all you’ll ever get is lousy noise removal smeared JPG files.
About the only thing good you will get from the pixel binning is perhaps better video quality.
Hank
4 months ago |Maybe you should do some research before you buy…. You know, so you know what you are buying and therefor can’t be complaining like you are now:-)
Sahaja
4 months ago |@PhotoNut >”Everything was “auto” mode. I was so p*ssed off and swore off buying any Sony compacts again.”
Don’t you check out the specs *before* you buy a camera? Why be p*ssed off at Sony if they didn’t advertise that the camera had other modes?
Tom
4 months ago |Or… if you’re feeling particularly daring, try one in a shop! You know, in the real World, works for me.
Alas he read about it on the web and bought it on the web, be angry at the internet!
Matt
4 months ago |So effectively a quarter of the resolution? Couldn’t the same thing be done in post processing more effectively?
SonyA77
4 months ago |Most compact camera users only know how to press the shutter button and upload to Facebook.
In fact I’d say the same for most dSLR users too… 
But yes, I agree. For the experienced user it would make more sense to capture the image at full resolution and then PP it. Indeed in PP it may be found that “binning” wasn’t necessary and a few NR tweaks are all that’s required.
Don Cox
4 months ago |Binning has the advantage of giving much smaller file sizes.
On a high definition camera (for example, 80 Megapixels), one might use binning for all routine shots, giving a usual 20 Mpix, and have the option to use all the pixels for jobs that need them.
For example, recording a large painting or tapestry, or shooting a landscape for a mural-size print, could use every pixel.
shamb
4 months ago |‘So effectively a quarter of the resolution? Couldn’t the same thing be done in post processing more effectively?’
No: you lose accuracy bcause you have more than x4 the noise. If you bin at the sensor, you combine *before* the ADC, so the signal is stronger (less s/n because you are only getting one ADC channels worth of noise, and because ADCs give worse s/n for small signals, its actually better than 1/4 of the noise). If you bin anytime after the ADC (such as in PP) then you have added much more noise than pixel binning before the ADC.
pancanikonpus
4 months ago |few questions:
1) pixel binning for camera alr exist on the market such a77, a65, nex7 is it still can be done via firmware? or it is need hardware change?
2) if summer compact camera do have this feature, what it mean to existing SLT a77, a65, nex7 owners about their camera?
3) how deep the impact of this pixel binning to a camera performance? (since 5dmk2 does, its seem critical!?)
other questions:
1) will sony going to fix the 1.8x crop video recording on a77 issue via firmware?
2) tenthering when?
3) will sony unlock ISO setting >1600 when in video recording?
thank you!
SonyA77
4 months ago |1) You don’t need in-camera pixel binning, do it yourself in PP.
2) I don’t see why A77 users should be concerned about compacts.
3) You’d have to ask 5DMkIII users.
other questions:
1) Doubt it. The crop is there for electronic image stabilisation. I don’t think they’ll provide a way of switching it off, but it would be nice.
2) Don’t hold your breath for tethering. It may never happen. I have to also use a Canon (long story) because the A77 doesn’t have it!
3) According to Sony, the 1600 ISO video limit is to prevent the sensor from overheating due to more power being drawn by the sensor at higher ISO. So I doubt it will be changed.
pancanikonpus
4 months ago |Thank you for educate me. Sound like bad news about to ask sony deliver such fixes in crop, iso, and tenthering.
which mean if today i bought flycam 3000 for stabilizer, yet i still cannot off the software stabilizer and suffer for 1.8x crop. and for the ISO, canon 600D able to use auto ISO in video mode but sony cant do… sound silly!
Carl
4 months ago |1. Pixel binning has to be done in hardware, so it can’t be retroactively added to a camera.
2. They’ll carry on using it like they always have.
3. It’ll mainly serve video modes (as 1080p is only two megapixels anyway). The better one is at post processing, the less affect it’ll have on still photography, though it should have a noticeable affect on in camera JPEGs at lower resolutions and high ISOs.
monkeyfacemcbride
4 months ago |@Carl
Agreed. As I said in my post above, it needs to be done before the ADCs.
pancanikonpus
4 months ago |Thank you educate me, Carl.
chris
4 months ago |No Canon DSLR can do Pixel Binning!!!!
admin
4 months ago |Sorry corrected. The rumored 5D markII (not markII) will do it!
fotowolf61
4 months ago |You mean 5DMIII I guess.
BTW where is the new A77 FW 2.0 ??????
It is really needed to fix many stupid software bugs.
PhotoNut
4 months ago |BS. The Canon 5D MkII already does one axis pixel binning in video mode.
Fongios
4 months ago |What is wrong with the video crop i cannot really understand the problem
monkeyfacemcbride
4 months ago |Its a pain if you want to use a wide angle lens. For instance a 12mm FF lens which would normally be 18mm on APSC, becomes 22mm. An 18mm kit lens goes from an acceptable 27mm to a more mundane 34mm. Its great if you do telephoto stuff but if you want to video a nice scenic view it can be annoying.
SonyA77
4 months ago |It’s a real pain when you totally forget about the crop and “crop” stuff off, grrr!
monkeyfacemcbride
4 months ago |lol, yep, here’s one that happens to me all too often. I get a bird or animal in the frame, zoom in to maximize the frame usage, the shot looks wonderful, hit the video button and half the critter gets cropped out. Stop video, zoom out, hit video button….still too cropped. Stop, zoom out…and critter gets bored and wanders/flies off. It can try your patience a bit
Matt
4 months ago |Can’t you just put it in video mode on the mode dial that shows the crop beforehand?
monkeyfacemcbride
4 months ago |That would be the way to do it, but sometimes you just need to make the quick swap to shoot video without a visit to the “cat screen”. Ideally, it would be nice if the camera could remember your choice of video mode (PASM) and you could toggle between that video mode and your chosen photo mode through a custom button and it would take care of any required AF/MF switching etc.
SonyA77
4 months ago |@Matt
But that would mean taking my A77 dial off AUTO….
Ray Oody
4 months ago |So the tx200v will have it?? And hey admin did I dream of a new HX cybershot or there will really be one ? Lol r we going to hear about the new summer compact cameras by 13 march ?
C
4 months ago |As long as the entire sensor area is binned (aka every pixel is summed) then I look forward to the video mode.
Cliff
4 months ago |Hmmm….the diagram doesn’t illustrate how it reads out and combines the RGB Bayer pattern.
I also wonder how the mathematical output is compared to a “true”, single large pixel.
Very cool. I have also heard of this referred to as a “super pixel” a couple years ago in another forum.
monkeyfacemcbride
4 months ago |AFAIK You use a modified bayer mask. In a conventional bayer mask all Greens are the same all, blues are the same etc. In the bayer mask for a camera that bins you alter it such that the densities of the RGB filters are not identical for all three components in any adjacent metapixels. A colour calculator knows what the exact mapping of the filter is and when receiving the luminance result for a given metapixel knows what proportion of the light comes from which colour component. It then does some complicated interpolation with the surrounding metapixels, comparing the component values from those to ascertain the true value for each of the metapixels….its not particularly straight forward and implementations probably differ.
the mathematical output is probably not quite as good as a single large pixel, but a lot better than 4 combined in post. This is more about versatility then exactness
Super-pixel or meta-pixel are the terms I’m familiar with.
Steve Jones
4 months ago |This needs a bit more about the technology than a quick cut-and-paste job out of Wikipedia. There’s no technical issue with pixel binning B&W cameras (some B&W CCTV sensor chips do it). The problem is with colour. On a Bayer matrix sensor (and any other likely colour filter matrix), adjacent pixels are not for the same colour channel, so can’t simply bin 4 adjacent pixels to get a “super pixel” as it would be an uncomfortable mix of 2 x green, 1 x blue and 1 x red with no colour separation. Binning the next nearest pixel of the same colour would mean skipping pixels. Do that and it will have a much more serious effect on resolution. To get 4 blue or red pixels on a Bayer pattern matrix, you’d have to reduce area resolution by a factor of 16 (as you’d need a 4 x 4 matrix to get 4 blue, 4 red & 8 green pixels). So a 16MP sensor would spit out a 1MP image.
Of course this isn’t a problem with a Foveon-style chip, which samples all colour channels at all pixel positions. Possibly there’s some non-standard filter matrix that would not have to do this for non-Foveon, but if so, I can’t work out what it is.
In any case, if you have a good ADC converter with low read noise, you can achieve essentially the same thing through down-sizing anyway (and more flexibly).
I know there’s an MF sensor that claims to do this, but there’s absolutely no technical detail about how it works, so the suspicion must be it’s downsizing as I seem to recall it outputs a TIFF.
Incidentally, the diagram in this item looks like a CCD, not CMOS sensor.
saturazzi
4 months ago |Wouldn’t binning of four colour pixels (2 green + 1 blue + 1 red) give some kind of one simulated “Foveon pixel”? Also, 2×2 binning of a, let’s say 24 Mpix sensor, would give a better resolution than a normal 6 Mpix sensor, because the low pass filter would have much less negative impact on the resolution.
Cliff
4 months ago |Is it CCD because its reading and binning from two directions, bottom to top and then right?
Could it be a CMOS with a global shutter?
This is a great topic!
monkeyfacemcbride
4 months ago |It’ll be CMOS probably CCDs are a bit 90s now due to power consumption issues. Basically you use parallel readouts (2 rows/columns at a time) feeding into a single summing well. Clock out 2 pixels from 2 columns/rows (so 4 pixels in all) into the summing well where the charges accumulate, then clock the summing well out to the readout channel where its amped and ADC’d.
Steve Jones
4 months ago |The original post had some diagrams that were most certainly CCD as it showed the “bucket chain” type reading. However, those have now gone. I still don’t see how pixel binning could work on a bayer matrix style sensor.
Incidentally, CCDs are still very popular on compact cameras. DPReview shows no less than 337 of them, far outnumbering CMOS on compacts (only about 80).
monkeyfacemcbride
4 months ago |As I said in an earlier post, you dont use the standard bayer matrix we all know where the densities of all green filters are equal, all blues are equal etc. You use a modified one with uneven density filters. In addition to this you have a heavyweight colour calculation unit which knows which filters of which densities occur in which metapixels. Then you use a series of interpolation patterns to ascertain the true metapixel value – basically you do 4 or more passes over adjacent metapixels.
For example all metapixels would have a total max density value of say 3. within one metapixel you may have one G pixel with a filter density of 1 an R with a density of 1 and a B and another G pixel with a densities of .5 (making 3 in all). these RGB ratios vary from metapixel to metapixel such that no two adjacent metapixels have the same combination of densities. The Colour calculation unit knows the mapping and can ascertain from the luminance biases of adjacent metapixels what the ratio of RGB for each metapixel should be. As I said its not straightforward.
Power consumption is a function of size for sensors so CCD is fine for smaller sensors (compacts, security cams etc), not so good for full frame DSLRs. Early DSLRs did have CCDs but that was more because CMOS tech was a little immature at the time. When it was perfected CMOS ran away with it.
Steve Jones
4 months ago |“heavyweight colour calculation” does not sound like binning. Binning involves adding charges. What you describe is a complex version of downsizing based on a distinctly odd layout of filters. I can’t see how what you describe will work – if you have filters of different density, the lower density ones will saturate before the higher density ones (with white light). That’s distinctly sub-optimal as the high-density ones will be under-exposed with poor SNR.
So what you describe isn’t binning. This is one site that purports to describe how the Phase-1 sensor performs “pixel binning” without making complete sense as far as I can see.
Randy S.
4 months ago |Which is why all highend camera are all CCD still? Phase One. Leaf, Jenoptiks. ALL CCD.
monkeyfacemcbride
4 months ago |@Steve Jones
“Binning involves adding charges” – Yes, I know I’ve said that 3 times the rest of my responses to the article.
“So what you describe isn’t binning.” – Yes it is, no ifs, no buts, it is – FACT, get used to it. Its a method of binning and a very versatile one at that, so much so that Panasonic patented it on the 29th of September last year. So if you think it doesn’t work then I suggest you write a quick letter to Fumio Ohtsubo (president of panasonic global) at 1006, Oaza Kadoma, Kadoma-shi, Osaka 571-8501, Japan and tell him that Inokuma Kazuyuki (one of their best sensor guys) is far less smart than you and should be fired…ok? BTW he’ll probably want to know a technical reason for your opinion, so best get your thinking cap on.
– Your ideas of binning sound a little shakey, so perhaps best not to be too judgemental of other peoples methods eh? All sensors, binning or otherwise have to do colour calculation to extrapolate the non-native colour components for each pixel. In this particular method its slightly more onerous as there’s more to this method than there is to most other binning methods and as I said in an earlier post, this is just one implementation there are plenty to choose from, but most that I’ve seen involve non-standard bayer masks and “heavyweight colour calculation”.
monkeyfacemcbride
4 months ago |@Randy S
MF back manufacturers using CCDs, well might this be because:
A. They dont care about power consumption – average 200-300 shots per 2000mAh battery (pentax 645 excepted but this is a different case for reasons I cant be bothered to explain). Most MF cameras are used in studios not the middle of nowhere like DSLRs
B. CCDs have better dynamic range than CMOS
C. CCDs are less complex to design than CMOS
D. CCDs are less complex to manufacture and generally have higher yields than CMOS which is important in low volume production.
E. Producers of medium format CMOS sensors aren’t exactly 2 a penny. For that matter, there’s only really a couple of players in the MF CCD sensor market
LEAF incidentally were producing medium format CMOS backs as far back as 2003.
Spoon
4 months ago |Actually, the latest CMOS sensors have surpassed CCD’s in their DR. Lower read noise being one reason.
The last three points mentionted, cover it mostly.
monkeyfacemcbride
4 months ago |@Spoon
You’re quite probably correct, for that one I was going on comments from Phase 1 a couple of years back.
Oh and in addition, remember the FZ150 vs D3S issue from a while back, after much research following your response to my question, you were also correct on that too.
skeptical
4 months ago |What are you talking about? You take a 4×4 matrix and extract from that 2 aggregate green pixels, 1 red and 1 blue. Originally, the 4×4 matrix was 8 green pixels, 4 red and 4 blue. So it’s a reduction by a factor of 4, in each color. No difference between color and B&W. What is your point?
Chris
4 months ago |I didn’t know people expect to print wall sized photos with P&S.
Just make sensible mp cameras, for God’s sake.
Cliff
4 months ago |+1
rovhazman
4 months ago |The way I understand how digital photography works (and I might be totally wrong…) does not allow real pixel binning!
Since we have the filters for different colors, neighboring pixels does not record the same data. So it wouldn’t work for RAW files and it will not be real binning.
Binning is done in scientific cameras, but they record B&W.
SonyA77
4 months ago |This is the stuff of people with beards and big foreheads, smoking pipes and wearing white coats. I’ll just take some photographs instead…
Francisco G
4 months ago |Exactly! hahah
MCP
4 months ago |Pixel binning is only new to Sony, they are just finally doing something others have done for a lot longer.
Copycats making it seem like they are innovators in the industry- just like the semi-transparent mirror, waterproof cameras, DSLR live view w/phase detect (oly did it first)- I could keep on going.
The only innovated division Sony had was cybershot with a couple of totally unique features.
end of line.
Steve Jones
4 months ago |Nonsense. Phase 1 apart (a special case), give the model name of any other colour still camera performing pixel binning. Just one.
PhotoNut
4 months ago |The Canon 5D MkII does single axis pixel binning when shooting video.
The Fuji EXR type cameras also do pixel binning.
MCP
4 months ago |And there are Steve. The only nonsense is your reply without doing your homework. And, that was only one of the things that Sony copy others doing that I mentioned. Quit defending before you do you research. No reason for it.
end of line.
Holy Bear
4 months ago |“Pixel binning is only new to Sony”…..I don’t think Samsung/Panasonic/CASIO/NIKON/OLYMPUS’s cameras can do this….
MCP
4 months ago |Better check again on Nikon. And still that leaves several manufacturers that did it before Sony did it- nothing innovative, just copying.
end of line.
Gabriel
4 months ago |I don’t know if it’s a true pixel binning, but the new nikon 7100 can reduce noise buy taking picture at a 3mpx with some pixel combination.
SJP
4 months ago |The post isn’t about it being new technology that Sony invented – its about Sony using it in new cameras. People on here don’t have to get their facts straight – you just need to read the post correctly.
SonyA77
4 months ago |Leave MCP alone, he has special needs.
end of line.
MCP
4 months ago |Copy cats that don’t do it better than the ones they copy cat- they suck.
end of line.
Spoon
4 months ago |With read noise down to 2 to 3 e-, I don’t see much use for it in Sony’s large sensor cameras when shooting stills. Unless you somehow are willing to give up lots of resolution to have smaller files.
For video, there could be benefits, if they somehow pull off true binning.
Walt
4 months ago |4 pixel combining would just be what’s done to obtain full color from the monochrome pixels of the sensor. So a lot more pixels would have to be combined to be any different than now is done just to get color. And a lot more resolution loss.
knurd
4 months ago |I seriously hope this is the next step for sensors from all companies. I don’t want my next camera to be more mp at larger file sizes but same quality output. The majority of people don’t print photos larger than 8×10 on a regular basis – if at all. If anyone needs to print larger sizes then there are already great cameras for that or they can hire a pro.
Qphoto
4 months ago |I’m not sure what all the fuss around “pixel binning” is, except that this is the first I’ve seen in a compact cam. It’s has been around for years and is already in practice on PhaseOne MF backs P40+, P65+ and all the IQ series backs. This is what allows them to perform better high ISO at lower resolutions, thus increase speed, sensor sensitivity and reduce noise out of those higher MP sensors (with the loss of mp to gain that performance). I see the same thing from the compacts with high MP, IMHO.
There should be some technical documents on their site of floating around the next if you want to do a deep dive or head over to LL or Getdpi and do a search to find old posts for its implementation on CCD sensors.
Q
monkeyfacemcbride
4 months ago |Panasonic and Fuji have been using it in their compacts for a couple of years now.
mark
3 months ago |…and it looks like Nokia will also use this technology in their 808 smart phone.
Street Wizard
4 months ago |Read about pixel binning here: http://www.echenique.com/index.php/2009/05/16/phaseone-sensor-pixel-binning-and-the-future-of-digital-photography/