New Panasonic “Micro Color Splitters” sensor tech explained on video.


The video on top explains the tech behind the new innovative Panasonic “Micro Color Splitters” sensor. Panasonic said that new technique allows to “approximately double the color sensitivity in comparison with conventional sensors that use color filters.“.

I hope we will not have to wait to wait a long time to see it on use on a Micro Four Thirds camera!

P.S.: Thanks to the anonymous sender and to Tim for sending the link!

  • mw

    pretty cool

  • Looks pretty cool.

  • Pat

    Panasonic hasn’t really had a chip breakthough (or even a leading chip) for sometime, this will really shake things up.

  • Anonymous

    Yeah but Olympus has IBS! Ha ha ha ha. They’r’re going to need it.

  • espen-b

    Start saving for the gh4

    • V4Vendetta

      GH5, 4 is is a bad luck number in Japan.

    • V4Vendetta

      GH5, the 4 its a bad luck number in Japan.

      • I always hear people saying that, but how about Nikon F4, D4, Canon G4 or Olympus OM-4.

        • digifan

          And 400D, D400, FZ40 etc. etc.

        • but OM-4 was the last OM, so in a way bad luck after all ;)

  • Sleeper

    Nice fine, that’s very cool.

  • donsantos

    This would make micro 4/3rd challenge full frame cameras in terms of low light performance.

    • dac38555

      “This would make micro 4/3rd challenge full frame cameras in terms of low light performance.”
      Hahahahah Bahahahahhahahhahahahhahaha

      • true homer

        Care to explain with a logical statement whats so funny?

        • Carlos

          The same technology can be used in a full frame sensor which would make it superior than the 4/3. The great thing about the m4/3 isthat you can have small lenses! So portability is great.

          • Only if the new technology can cost effectively be produced at a larger size and only if Panasonic sells the sensors. Remember Fuji and Foveon are exclusive sensor types.


            • observer

              Remember BSI and SmartFSI are technologies that reduces or eliminates the disadvantages of the small pixel size sensor with respect to larger pixel size sensor. So it is most likely in future there will be little difference between small sensors and big sensors in light capturing as their Quantum efficiences difference would be negligible.

              • NFT

                please panny use SmartFSI in new G/GF series,
                SmartFSI is like more capturing(absorb) in same pixel size,they (panny) assume light income to sensor by several angels,SmartFSI will capture all or most.
                if you can show more detail ,thanks

              • Esa Tuunanen

                Signal strength given by different size pixels will always be limited by area of pixel and how much light/photons it receives.
                Quantum efficiency has nothing to do with that but only about how big part of photons hitting photosite cause change in charge.

                • Anonymous

                  Don’t misconstrue what I said. There is a diffience but the difference will be negligible. Keyword here is Negligible. If small sensors can give noise free images say at ISO12800 does it matter or necessary to go higher for normal everyday picture taking?

                  • Tom

                    Full frame with bigger pixels will have higher dynamic range, improving sensitivity or efficiency does not increase dr, which will always be below the dynamic range of the natural world.

                    • Jørgen

                      We are talking about sensortechnology here. So with that in mind and known differences between vendors, your comment is incorrect. The Canon FF sensors at base ISO are behind several APS-c and mFT sensors when it comes to DR. A technology like this, used by only one vendor could bridge the gap (in part) to current State of the art FF sensors.

                • NFT

                  @Esa,wrong to my comment??
                  thank for your comment.

                  but i announce pixel don’t need more area(bigger than) i think tech of microelectronic will solve old problem.
                  Assume ,i had old gf-1 sensor that bad DR and ISO when compare to epm2 sensor ,if i rebuild with SmartSFI
                  is better than om-d/gh-3 sensor ,
                  i don’t care about your factors but if tiny sensor can capture more or most light
                  i don’t want compare to FF sensor or FF + SmartFSI .it was great solution .

          • digifan

            Patents, patents.

            • true homer

              no no no, this is far beyond a patent. Panasonic UK said this will be seen INSIDE 2013

    • Jørgen

      My thoughts too. IF…..Let’s wait and see if this a viable technology.

  • This all looks promising. If its using standard fabrication methods it should not cost any more than current sensors. Does the splitting cause cross sensor bleeding and reduce sharpness or is the sharpness better because of more light to play with. I would like to see some samples taken with the 12-35mm and 25mm lenses. Real pictures not charts on walls and graphs. Use the graphs to explain the difference not the images.

    Where would we be if it was limited to m43 production only?

    And the ‘is it is, or is it ain’t” question. If the lens has diffraction is that good or bad for the sensor diffraction???? :D

    • It would be in Panys interests to make it m43 only for at least 2 years. It would give the m43 still and video a huge boost. Then go to pay for rights usage by others after m43 has made a set market target.
      It would also be in panys interest to let Oly use it as many oly users use pany lenses.

      • gtui

        It would be in Panasonic’s interest to keep the technology for their own camera range giving them a potential advanatge over Olympus. Panasonic and Olympus are not partners in the end they are rivals, with neither company doing financially well in their respective imaging divisions.

        • Jørgen

          I tend to disagree for various reasons:

          1) Panny sells lenses and they sell them for Oly bodies too. Oly gone means less lenssales because people will rethink their new cam system in part…

          2) Witholding technology is not a smart thing to do. More on that later. Blocking things is not smart at all in the long run. It is better to make use of it.
          Sony sells their sensors to Oly giving Oly a huge boost with OMD. Panny can do the same. They are not giving the sensors away for free.

          3) MFT is seens as a system that clearly consists of two players covering somewhat different fields. Oly being more stills directed. Bodies are smaller, no DSLR style. Strenght are prim lenses. Panny more video and DSLR directed, zooms (also for video) are strengths (X-series, 12-35 and 35-100).

          Combined this makes for quite a versatile system. No need to compete strongly. But in the past, Panny did just that. So I come back why witholding technology is not a smart thing to do: Oly might have jumped ship because of this. Now Panny sells very few sensors to Oly. May be the Sony was just too good as the GH3 uses it too, but that did not come up just liek that. Panny withheld the 16 MPixel sensors form Oly for almost two years. and now they are selling zero sensors to Oly…(Only in the old Oly cams).

          • Anonymous


            “Panny sells lenses and they sell them for Oly bodies too. Oly gone means less lenssales because people will rethink their new cam system in part…”

            Spot on, one of the main selling point for MFT, and what got me started, was that it was a relatively open system where vendors compete on the quality of their product rather than locking me into their system.

            I sometimes buy things from Olympus, other times from Panasonic and occasionally something more exotic from Voigtlaender and potentially SK in the future.

            If Olympus or Panasonic were operating proprietary system mounts I would probably not have gone with either and they wouldn’t have soled me a single thing.

  • I know Olympus is using Sony sensors now but I just wonder what could Olympus do with this sensor.

    • Well GH3 have same sensor so is in E-M5 and last Pen camera and have also patent so is from Olympus, so why can this not have a patent from Panasonic to.

      • PannySensor

        If the OMD and GH3 share the same sensor, you can best believe it’s a Panasonic sensor. Only Olympus fanboys think its a Sony sensor. Olympus fanboys love being lied to and paying extra for Lenses, and etc that are mostly standard for every other camera company.

        I don’t see Olympus lasting long. I see them being brought out by another company. I also believe that Panasonic would like to further distance themselves from Olympus, being that they sided with Sony. The heads at Olympus just keep making mistake after mistake with just a few highlights (OMD) in between.

        There is no way Olympus gets their hands on this new sensor. Not while they’re in bed with Sony. If I was Panasonic, TM, patent or not, there is no way I would pass off new tech to Olympus, knowing they’re partners with Sony, a rival company.

        • Not Olympus fanboys but Sony fanboys :-P
          I think Sony not have made FT sensor, and not will do this too. :-D
          But Olympus use Sony sensor in many product, so in microscope series IX-3. ;-)

  • Renato S.

    It seems pretty viable now, the only thing that concerned me is when he talks about processing power needs. So I am assuming that the best you can do is to develop a processor that is optimized for this light splitting process, if not it will take a lot of processing power for this.

    Can Panasonic do that now? So far it seems better than the x-trans and maybe even foveon as a practical and improved sensor, I think that the ISO and DR will be pretty good and without the color filter, can I also assume it will be sharper as well?

    • Chez Wimpy

      I’m pretty sure he was talking about processing necessary to model the various designs (hours vs minutes) leading up to the final optics selection for the sensor. The actual image processing by said sensor (going by the equations there) should be trivial for any digital camera.

      • NFT

        yeah,to much processing need,i announce to look Smart Phone
        Circuit diagram will be image processor -> ARM Architect CPU ,Cpu can process like image processor .
        I mean the future digital cam need massive processing like CPU flt. OS.

  • Anonymous

    Looks interesting, and I’m cautiously optimistic, but remember that the only things we’re hearing are things Panasonic wants us to hear. Foveon didn’t talk about the drawbacks of their sensor, either.

  • observer

    Interesting that it has moved from the patent stage to fine tuning the early prototypes. Maybe a real product and mass production in 2 years?

    • I hope Panasonic will use this patent on next multi aspect sensor. ;-)

      • observer

        It should be a colour splitter, multi aspect sensor with global shutter. :)

  • Tron

    Panasonic engineers have gone rogue! This would definitely pimp smack FF sensors around if it could be implemented in the near term.

    • I think also FF guys not like this here, and if this patent will be coming on a FF sensor need them displacement microlens out in corner like Leica have on FF sensor. :-)

    • twoomy

      Will you guys stop being idiots about pimp-smacking, schooling, crushing, and killing full-frame just because of a possible sensor improvement? As an M43 shooter *and* a Nikon FX shooter, I’m happy for improvements in M43 image quality, but it’s not going to ever [insert random violent victorious sports term] FX cameras. There’s a market and advantage to both systems.

  • Anonymous

    Panasonic , dont just do innovation,.but you have to start put that in your camera A S A P,..

  • Anonymous

    You are going to get the same amount of light as the Leica Monochrome.

  • matt jones

    So you will end up with the same amount of light as the Leica Monochrome.

  • cab10886

    Well… lately I’ve been reading up on digital Infrared photography and full spectrum photography… Since they are using micro splitters that split different light frequencies… how would this affect IR light? Would the sensor still need the IR hot mirror or would the micro splitters end up blocking IR light? Would IR photography be possible on a sensor using this technology?

    • observer

      We should assume the colour splitter lens only split what’s visible light spectrum and hence a natural filter for infra-red or ultra-violet range.

    • Well Panasonic tell us not by them use lowpass and higtpass filter here, but reason for this patent is only the not have, and not can use Bayer filter on sensor.


    The concept behind this is great. However, one issue that is not discussed here is the behavior of the micro-splitters with non-perpendicular incident rays. It looks to me very possible that the splitting/diffraction behavior requires 90 degree incidence to work best. The cross-sections show an array of convex micro-lenses at the surface, which helps to reduce the effective angle of off-axis rays but cannot completely correct them.

    So, I am concerned that this type of sensor may show a high degree of color-decoding error with highly non-telecentric lenses (compact wide angle lenses in particular).

    This has always been an issue with digital sensors vs. film, though various developments have reduced the effect. Olympus specified the original 4/3 standard to require highly telecentric lenses that have near-perpendicular incidence. This was appropriate for high-performance digital imaging but also limits the compactness and flexibility of the lens designs. Micro 4/3 was able to relax the telecentricity requirement partly because newer sensors are coping better with off-axis rays.

    But now, if this color-splitter technology does in fact show serious artifacts with non-telecentric lenses, it may not be a completely problem-free improvement for M43 photography. Of course there could be other possible disadvantages that are not exposed in the company announcements – we’ll see. I’d love to be wrong, but usually there’s no free lunch. Maybe the patent itself discusses some of these issues.

    • observer

      From the diagram shown above, it seems that conventional colour separation by adsorption is only 30% to 50% transmission efficient as it has to factor in problems of oblique light and its colour leakages among other factors of direct light loss from colour filtration.(See also about SmartFSI )

      Once the colour filter is discarded and replaced by the colour splitter, the light transmission is 100%?

      So the oblique light problem is only inherent to the colour filter and not to the colour splitter sensor?

      • NFT

        why no man design several angles pixel for absorb light and color ,
        that i mean if pixel read rays in others angle ,don’t require 90 degree only.

      • NFT

        curve sensor is nonsense.

      • Esa Tuunanen

        You need to observe lot better.
        If anything Bayer sensor’s filter based colour separation should be lot less sensitive to incident angle of light than this which depends on certain part of light being deflected correctly by those deflectors.

        Transmission efficiency of colour filtering/separation is all about how much of light is let through to pixels of sensor.
        Bayer colour filter array simply allows green light through for half of the pixels (those green pixels in Bayer pattern) while blocking other half resulting 50% transmission efficiency for green light. And for blue and red colours it blocks 3/4 of light.

        100% transmission efficiency doesn’t mean much to final efficiency.
        Also Foveon has 100% transmission efficiency while having craptacular actual efficiency.
        Only because instead of actual usefull RGB data Foveon sensor vomits clump of total mess.

        • observer


          You need to read better.

          Sigma never PUBLISH the claim that they achieve 100% transmission. They claim to capture “all the colours” NOT “all the light”. The difference showed in their high ISOs images.

        • NFT

          accurate color ,but not mean accurate light(capture);

  • Yun

    You don’t need to wait too long to see the first MCS camera & probably in this year end .
    It will in a G series camera known as very very highend camera , the theory of Leica M camera from Pana looks
    on the card .

  • I don’t see the real advantage behind this concept:

    50% more light would be 1/2 stop and that can be easily achieved in other ways like reducing circuitry and increasing the photo site etc.

    While no Bayer filter is used the colours still have to be interpolated and due to optical effects the splitting is limited so that the resolution is affected like with an additional A/A filter.

    The Foveon sensor concept seems much more promising with equally no loss of light but accurate colour information

    • observer

      It is hard to see the advantage of this concept if you dismiss it off-hand without even trying to understand it.

      Its reported advantage in the video above is 2-3times more light NOT 50%.

      You also failed to understand the Foveon sensor(“The Foveon sensor concept seems much more promising with equally no loss of light but accurate colour information”).

      Read the review of the latest Foveon sensor:

      Quote from review:

      “Noise is quite intrusive (with corned beef) at ISO1600, which will be particularly unpleasant in skin tones although the noise does appear to affect some colours more than others, and detail suffers. Colour is lost at ISO3200 and ISO6400 – these highest settings are best avoided.”

      The ISO performance for an APS-C size sensor(that of the Foveon) is way behind the competition. Lost of colour at ISO3200 and above!!!

      • I am not dismissing it but figure 1 shows that colour filters loose 50% light in comparison to the splitter. That equals 1 f stop in light gathering. At the sane time the white sites will be overexposed with the risk if highlight clipping while the coloured sites have the issue of noise due to underexposure. Either a dual sensitivity sensor like the one of Fuji has to be coupled with the splitter or the exposure has to be reduced. The real gain is hard to estimate and might very well be just 1/2 f stop but I am happy to be proven wrong.

        Maybe we need two sensors as in the human eye where we have colour receptors for good lighting conditions and sensitive greyscale receptors for low lighting conditions and the colours have to be extrapolated again. We might also have to accept that there are physical limits to everything and just work with what is available to us…

        • Black and white movies can be colourised by matching part of the grey scale with colour. Computers then extrapolating that to the entire frame to create full colour. This is flawed in terms of excellent quality because of the variation in film stock and processing. But in a ‘gripping’ movie has been be of sufficient quality to be acceptable
          However if we have a black and white sensor with some colour pixels in strategic positions it would be possible for the colour pixels to be used to extrapolate colour across the black and white image (an assumption that the b&w pixel next to the colour one is a basis for extrapolation for example). As this is done on a constant performance from the sensor (not variable like film may be) it could be made to give very high quality. Higher ISO and sharpness could be achieved on sensor but higher processing power would be required.

          • I guess it still requires a lot of R&D but I agree that this could be a way forward. Any mobile phone outperforms the computing powers of the lunar missions, so processing power will be available soon

    • Esa Tuunanen

      Foveon is lot worser, by design:
      It doesn’t output any sense making real colour data but vomits diarrhic mess.
      After all that is cleaned up there’s left only 1/4th of Bayer sensor’s efficiency in converting light projected by lens to actual usefull RGB image… and actually only 1/8th in low light.

      Full per pixel resolution (no demosaicing) is only advantage Foveon has.

      This pixel level light splitting to RGB subpixels is only really major advance promising concept:

      • Esa, I agree that the efficacy of the Foveon sensor is nowhere near to what we would wish. But that is less an issue of the design but the inability to produce the colour sensitive photo sites in the desired manner. It could therefore very well be that the “inferior” colour splitter will produce better images but to me, apart from the increased computation demands, the difference in exposure of white and coloured photo reception sites has to be addressed to achieve a dynamic range competitive to traditional Bayer sensors.

  • anyone

    noticed Window$ XP and HP/Dell monitors? WFT? They don’t use my Eizo? ;)
    Yeap, once GH5 has mar sensor back and focus peaking (I’ve just bought SLR Magic 25/.95) I’m all in…

    • asdfjkl

      It bugs me more that they are not running the monitor in it’s native resolution. The display is scaling, potentially with an incorrect aspect ratio (5:4 vs 4:3) depending on the size. That looks like a 17″ or 19″/20″ with a native resoultion of 1280×1024 or 1600×1200. It looks like they’re running it in 1024×768.

      It doesn’t seem accurate for scientific/imaging uses to allow in-display scaling to occur.

  • This is mobile phone camera tech, people. Don’t get too excited. Most of the R&D money is going to mobiles these days, because that’s what sells more.

    • Narretz

      This is what I thought. This design might best work with a fixed lens, and small chips are simply cheaper to produce. So I’ll guess they’ll try it in mobile first.

      • It says can use current fabrication methods. So no extra cost and no size limitation over the existing.

  • Wondering about the possibilities to have it in GH5. (Cannot see real chances for getting it in G6/7)

  • aqasem

    I don’t know why. but this tech might be used in the small sensors for coming LX7 & FZ200

  • john

    All hail windows XP.
    So, if I got it right, those four cells will make up one pixel, containing all RBG values after interpolation, right?

    I guess they will hardly try to implement the color interpolation algorithm using standard general purpose CPU/GPUs. I don’t know anything about the algorithm used but often these kinds of algorithms scale very well when used with dedicated circuitry so I’m quite optimistic. Dedicated circuitry usually also decreases energy costs by an enormous factor.

    Raw conversion will probably become unnecessary because you’ll hardly if ever carry out the interpolation on your PC. However, one thing that bugs me is that they did not say that their new algorithm is mathematically equivalent to the old one. Since they don’t mention it, I assume there will be some drawbacks (in the video, the scientist was like, ‘oh, look at that, they look almost the same! Excellent!’), so people with high aspirations may still resort to raw conversion based on pure sensor data. I guess that would be nightmarish slow however…

  • Vivek

    Appreciate this post and the video link! Thanks!

  • Newbie Here

    There are soooooooooo many techies here, my head hurts.

  • maxter

    am I the only one that thinks that Panny techs using Oly microscopes is kind of funny…?

    • Liam

      Panasonic and Olympus are secretly the same company owned by a Canadian. :D

      • Is that an Adian living in the can?

    • Yes.
      Coopetition is the real business. Every one cooperates but are in competition at the same time.
      I make my gear in your factory and then sue you when you make yours look like mine (nothing personal should be implied here). but I’m not getting my gear made elsewhere, cause its all good publicity. The public loves one of us and hates the other. But it makes all the parts cheaper and reduces production costs and we keep anyone not in our coopetition out of the market.

By continuing to use the site, you agree to the use of cookies. more information

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.

What are Cookies?
A cookie is a small file of letters and numbers that is stored in a temporary location on your computer to allow our website to distinguish you from other users of the website. If you don't want to accept cookies, you'll still be able to browse the site and use it for research purposes. Most web browsers have cookies enabled, but at the bottom of this page you can see how to disable cookies. Please note that cookies can't harm your computer. We don't store personally identifiable information in the cookies, but we do use encrypted information gathered from them to help provide you with a good experience when you browse our website and also allow us to improve our site. You can watch a simple video from Google to find more information about cookies.

Cookies used by our Website
The 43rumors website,, uses the following cookies for the collection of website usage statistics and to ensure that we can . These are anonymous and temporary. By using our website, you agree that we may place these types of cookies on your device.
Read how Google uses data when you use our partners' sites or apps:
Google Analytics Cookie Usage on Websites: Addthis cookies:
Disqus cookies:
Vimeo cookies:
Youtube cookies:

Disabling/Enabling Cookies
You have the ability to accept or decline cookies by modifying the settings in your browser. Please note however that by deleting our cookies or disabling future cookies you may not be able to access certain areas or features of our site. For information about how to disable cookies in your browser please visit the About Cookies website.