More thoughts about the “new” Olympus sensor.


The real mystery about the upcoming new PEN’s is the new 12 Megapixel sensor. I have been told that Olympus markets it as a “completely new” sensor. And I know for sure that it is not going to be used by any future Panasonic camera. But I don’t know yet how deeply Olympus redesigned the sensor. There are basically two options:
1) The sensor is 100% from ground up new.
2) The basis is the same 12 Megapixel Panasonic G2 sensor and Olympus only did make some kind of changes on it.

The truth is that my sources tested the jpeg quality only so it’s difficult to say if the changes in image quality are the results of the sensor or of the new Truepic VI engine. The good news is that my sources noticed a visible improvement. Dynamic range has been increased (more in the dark areas than in the highlights) and so the colors (with better blue). They told me it’s the best Micro Four Thirds sensor to date. But let’s not get too enthusiastic. The improvement is here but not extraordinary. The best news is the fact that Olympus recognized the need to become independent from Panasonic sensors. Anyway, I am reposting a comment from Thom Hogan about the posisble future Oly sensor:

I don’t have the original source to look at for that statement, but…those three specific things say to me that this is a sensor redesign, not a from-scratch design. Let’s do it backwards:
* Autofocus speed: highly dependent upon how fast you get data off the sensor. There are two basic ways to do this: up the frame rate and sub-sample the pixels, preferably both. This is work mostly in the add-on circuitry in a CMOS sensor. It doesn’t touch the sensel itself (other than how fast it is taking information from it).
* Color fidelity: highly dependent upon the Bayer filtration. We’ve been all over the map on this one during digital’s history, increasing filtration, decreasing it, flipping it to CYMG instead of RGBG, you name it. It’s an optical issue, and Olympus is an optical company.
* Dynamic range: can be dependent upon a lot of things, including sensel efficiency, read noise, microlenses, and so on. But read this and the previous one together and I get “Olympus changed the layers on the top of the sensor.”
I’m guessing here, but it’s quite possible that Olympus really meant it when they said 12mp is enough (it isn’t, but for different reasons than most people think, and Lytro is a good example of that). With Panasonic showing them sensors with similar traits but more pixels, Olympus might have simply asked to be given access to the original sensor design for improvement. So they put engineers working on the two easiest things to tweak: the photosite support circuitry (goosing the bandwidth) and the filtration layers (Bayer, microlenses). This would also explain why Panasonic is the fab.

I’ll be watching the details in the Japanese press when the release happens. What I’m looking for is evidence that the sensel is changed or different, or that the underlying transistor type is changed. Differences in substrate materials or flipping the sensor (BSI) are also fundamental design changes. Those are all things that would indicate a ground-up sensor design. But based upon what we’re hearing as leaks, I’d bet this is a reconfiguration of an existing sensor.

  • @admin
    preferred it when there was 50 comments to a post, instead of endless scrolling and slower loading times :(

    • with the recent 300+ comment threads youll have to browse thru 6 different pages, i think itll be much slower then the new method

    • Now is better!
      Txt is not a heavy load and is easy rendered by the browser.

    • Neonart

      The loading issue is the big ad on the top. I disabled Jacascript on my old G4 I use at work, and that fixed it. On newer machines it doesn’t seem to be as much of an issue.

      @Admin. That top ad is not very efficient. Can it be streamlined? Great work though! Not trying to criticize, just help.

      • admin

        Sorry, whch top Ad? There is no top ad here.

        • Neonart

          The one with the photos of each camera.

          • Maley

            Thats not an ad…

            • Neonart

              Admin, it’s the “Camera database” with the photos of each camera. Did a quick test with Javascript enabled/disable on an older machine. With javascript enabled the site takes an average of 14.15 seconds to load (if no warnings come out stalling completely ), with it disabled loading averages in 6.73 seconds and no warnings at all. This is new to the site redesign.

              Maley, if there are photos of products with four links under each to direct you to a place to purchase thats called an ad.

              I don’t have any issues with ads or links, just that that one seems to be buggy on older machines. I’m Just trying to help. It doesn’t slow down my MBP enough to be problem, so I leave it on.

    • beomagi

      My blackberry opens this fine…

      Better than loading a new page over and over.

    • XA4

      I like the new format, only one or two pages.

  • I just found this in the dpreview forum:

    Any comments on that?

    • napalm

      that’s interesting

    • Bob B.

      The referred posts are technically over my head!?.LOL..does that mean that there is no IR filter in front of the sensor like on a Leica M9???

    • I’d love it if this was true.
      With a predictive AF working fast and accurately, it would be farewell FT and hello again µFT. That’s what I’ve been waiting for since I switched from the G1 to the E-30 and then E-5.
      For me – I’m shooting mainly dogs and other animals in action – this would be the greatest news since I left my P&S days behind and began regarding photography as a real hobby.

      Seems like Thom Hogan might be wrong, this time.

      • Patric

        Hope they will have the same system with an E-5 successor. The bad thing is that I have invested quite a lot of money in FT lenses, and they will probably not benefit so much (if anything?) from this system. The E-3 and E-5 body ergonomics are perfect for me, so I don’t want to go the mFT way.

        • Disraeli

          Patric FT lenses still transmit IR light, and old FT sensors have an IR filter. I can be convinced they will not be optimum for the new system, but not the same. We dont know enough about how this works yet to say, but it is more usual for improvements to grow across the board.

      • Disraeli

        I think I would hold my breath a little on that Don, I do not think Thom is wrong at all.

      • I’d be happy to be proved wrong. I think it should be clear that I was speculating.

        As for comparing IR to non-IR, that certainly is one of the techniques that was being explored by a number of organizations. The problem with it is that if IR gets to the imaging sensor you have issues with color pollution. A lot of cameras have been burned by this, including the original Kodak DSC-2000 and the Nikon D2h, to mention to very big ones.

        It’ll be easy enough to test whether IR data is causing the focus speed changes, so we’ll see when the cameras get here.

        • Hi Thom, my comment wasn’t ment as an offense.
          I just would be virtually over the moon if they would sueprise us with an µFT camera with working C-AF.
          Every single day, I’m schlepping the E-5 and four lenses around with me for several hours while walking my dogs.
          And all the time, thevtitle of the old Gangstarr tune ‘Who’s gonna takre the weight’ is going around my head.
          So , please understand that I didn’t want to criticize but just am extremely excited.

        • Thom, couldn’t you break the spectrum at the microlens level?

          • @donparrot: didn’t take it as offensive, just trying to point out that I was speculating, not stating facts

            @sneye: sure, but that’s not apparently what they’re doing, see next

            @brian: Yeah, I saw it. But that’s giving up a G photosite for focus information. I hope we’re not losing the Bayer sensor’s efficiency for resolution. This matters more in low light. But it could explain one of the differences I saw in some early E-P3 shots.

          • Disraeli

            Yes thank you Brian

        • Leica M8.

        • Disraeli

          ah see, now thst likely wrong Thom
          it doesnt need to process an IR image, just shutter the IR cutoff filter off the sensor. Dont ask me how they do that, but it sounds like thats what theyre doing. What I commented on was your ideas of dual processing and multiple pipelines, I think thats going to have to be a certainty no matter how they do it

        • if the color gets affected, wouldn’t it be the easiest way to correct it within the image processor? I mean, i used a Kodak DCS520 without the removable AA/IR filter(the previous owner couldn’t find it anymore…) and the magenta cast was most of the time easily removed with auto white balance in post processing…otherwise the magenta cast gave the photos a look like they were scanned from old faded colorprints ;-)

    • But wasn’t it the special filter on the sensor which was cutting the IR off? Why lens coating?

      If they have moved IR filter from the camera to the lens, it might make all other m43 lenses sort’a incompatible.

      • Bob B.

        If there is no filter in front of the sensor then every image from the lenses can be that much sharper. One less layer to go thru….but Leica does that on the M9 without a special coating on the lenses…as I guess the M9 does not use infared to focus…not sure.

        • Disraeli

          sorry Bob, Leica use lenses as old as 60 years, they dont do anything special on lenses.

          The difference between the M8 and the M9 is the thickness and strength of the IR filter on the sensor, where M8 (not M9) requires an IR-cut filter on the lens. Leica applied a little special trickery on M9 as to the location of the filter within the sensor filter stack, thats all.

          Perhaps what you are thinking of is the anti-aliasing filter, which neither M8 or M9 have, and have a special sharpness due to that.

          • Bob B.

            Maybe it is the anti-aliasing filter I am thinking of…I am old and it is hard for me to retain meaningful info!!!! LOL!

            • Walter

              The antialiasing filter is indeed what’s going on, but it’s there for a reason. Look at a M9 jpeg sometime at 100% — they’re actually junk.

      • Disraeli

        Dummy, photographically the results would be very similar. There is little difference between having an IR-cut filter on the lens, or on the sensor.

      • Should this really be the case, it wouldn’t be a problem – if Olympus wanted. We could just send our FT Zuikos or older µFT Zuikos in to have them recoated – or let them replace the one lens with the coating. As I said: If Olympus wanted. I think this should be possible to do for a reasonably small amount of money.

        • That’s a really expensive proposition, probably not worth it for anything less than SHG lenses.

    • That will explain why they replacing the kit lens with “Infra Red” kit lens :0

  • The AF speed is amazing. Can’t believe they’ve managed to beat Panasonic on one of their strongest suits. Also good to see the sensor stay at 12MP instead of going to 18MP like the GH2. If it’s anything like ‘new’, in terms of the CMOS tech inside, that will be a great advantage for low light and dynamic range!!

  • Gasper

    14-42mm IIR as IR – Infra Red Focus

  • zuzu

    Take a look at this patent.
    This is indeed a new sensor (visible + infrared)
    New very interesting possibilities!

    • Very interesting, yes.

      • Marq

        Based on the patent link, it seems that the sensor has been fundamentally redesigned to take IR from the parts of the sensor aka Fujifilm style.

        • AndyOz

          Forgive my lack of knowledge – is this similar technology to that developed by Fuji which I understood would allow for on-sensor phase detection? It will be interesting to see how FT lenses focus – whether there is any improvement.

          • Same idea: sacrifice Bayer photosites for photosites that do something else. However, in Fujifilm’s idea they put in real phase detect AF sensing, in Olympus’s they use a different filter over the photosite to allow mostly (only?) IR to get through and be captured. Fujifilm also scattered their sites across the sensor while Olympus appears to be taking one of the Green positions in the Bayer filtration (i.e. putting focus at every RGB collection point of the sensor).

            • AndyOz

              Thom – thanks for the explanation, much appreciated. I would have thought that one of Olympus’ main drivers for this sort of technology would have been for the improved AF of existing 43 lenses. I wonder if their approach allows for such an improvement? Or whether its just new MFT lenses.
              It sounds like the Fuji phase system would have definetely improved the AF speed of 43 lenses but I guess Olympus wasnt able to use the exact same technology. You would have thought Oly would have let them join MFT and said as long as we can license your on sensor phase detection!

              • Depends upon whether the lenses have high levels of IR coating on them. Can’t speak to that for Oly, but Nikon doesn’t.

                My expectation would be that they did multiple things: sensor with the IR direction detection, revised AF calculation system maybe even using dual ASICs on the higher-end cameras, lowered IR filtration on the lens, plus a change to how the focus cams in the lenses themselves work. Like phase detection, the IR comparison should give you both a focus direction and a wide guess at distance (not a very specific guess ala PD).

                Thus, off the top of my head you might be able to do the same thing PD does: immediately move the lens the right direction in a big leap, then use CD to fine tune at the end. But older lenses aren’t made that way, they are designed to do small CD increments. Even if you can’t do that you could do a couple of wild leaps to establish where to concentrate CD movement. So, yes, a redesigned lens should be faster than an older, but older lenses should be faster with the new cameras than the old. Best of all worlds, really.

                • I remember – although I can’t find the link any more – that Olympus had a PDAF-on-sensor-concept patent already back in 2004 or 2005. So, they wouldn’t have needed Fuji to implement this technology. Therefore, I would assume that they came to the conclusion that the IR approch is more promising.

                • lily

                  “Depends upon whether the lenses have high levels of IR coating on them. Can’t speak to that for Oly, but Nikon doesn’t. ”

                  I can’t imagine that they would, seeing as how there are Infrared-Converted 4/3 cameras out there – there’s a 410 and a 510 on ebay right now. They wouldn’t work at all if the lenses were IR coated, right?

            • Ross

              The rumours are saying 35 AF points. Wouldn’t that mean only 35 0f the Green sites (or clusters of them) would be used?

  • Anton Rahmadi

    Hi, I emailed 43rumors admin just after my post in (my Indonesian friend’s owned website). Original post at:

    Visible+IR sensor is NOT new in medical science. I’ve been using it since a long time ago by shifting a filter. And Olympus may have an automated apparatus somewhere between microlens-sensor. Remember Fuji applied ND filter in-camera?

    To be frank, what ever technology Olympus going to bring on the coming Thursday, I’ll be happy to see where Olympus is up to with either micro4/3 or 4/3.

    Updated: My other friend has working sample of E-P3 and I am working with ‘this friend’ to test the AF and to confirm about the technology behind the camera.

    • OlyFan

      Very Interesting! Thanks for sharing this with us Anton.
      Please post your initial impressions of EP-3 here. I’m sure all of us here are waiting to hear more about it. terima kasih!

      • Anton Rahmadi

        Initial impression?
        Ugh.. “my friend” said: AF works amazingly fast on new lenses, but slightly better on older lenses. “this source” is bound to contract, could not pass more information.

        • Anton, any chance you can say something about the top shutterspeed?

          • Can’t tell. But ‘my mate’ is having a real excitement with this camera. I am sleepless now (in Aussie it is almost midnight!)

            • Anton, I can imagine that…i guess all of us here would like to test that camera :-)

          • Can’t tell. But my mate is having a real excitement with this camera. I am sleepless now (in Aussie it is almost midnight!)

            • Ross

              Yep & you must be too (or just getting tired) posting twice. ;)

              It seems like we have something good to look forward to on Thursday (about 3 to 4pm in Sydney?).

              • Anton Rahmadi

                The most interesting part is different region has its own rights to decide when the launching time is.
                Major descriptions have been circulated to distributors last week, but they are bound to NDA (non-disclosure-agreement).

  • Agent00soul

    Couldn’t it be the old sensor but with gapless microlenses and some other improvements?

  • Maciek

    There is a reason why the IR filter is used

    • An0n

      Very interesting. But if the sensor is processing both images — one with the iR, and one without, it stands to reason, the one without iR would be the one recorded. I’m sure someone else here with more technical knowledge can speak to this in more detail. But it seems Oly may have something here which works just fine in practical use.

  • cbr09

    Whether it works on 4/3 lenses will depend on whether the special coating is needed to get a specific IR wavelength or if they have to remove IR filtration which is there in existing lenses.

    If it is the former then a special filter might allow this work on existing lenses (could be included in the 4/3 to m4/3 adapter) but if existing lenses block these wavelengths you’d be back to existing algorithms.

    Anyway this is a really nice idea – and makes a lot of sense. It is also easy to see how they could quickly release updated lenses as they are just the same but with different coatings.

  • cbr09

    In principle if this is correct the cameras should also be capable of giving IR images – which would be interesting.

  • cbr09

    In principle if this is correct the cameras should also be capable of giving IR images – which would be interesting…

  • cbr09

    In principle, if this is correct, the cameras should also be capable of giving IR images – which would be interesting…

    Also this means the AF assist is presumably visible and IR – though they could allow a slower option of IR only which would be less obtrusive.

    • cbr09

      Sorry admin, I was getting a failure message on the post!

    • lily

      I would for sure buy one if IR images were an option.

  • Maciek

    most coating in lenses is against reflections and not IR, so it might work with 4/3

  • I may post something about ‘noise’. I am still verifying the information and digging deeply into technical details as if I am doing my lab research. It is coded as ‘noise prediction engine’

    • So they want to “predict” focus AND noise? WTF ? :-D
      But the 2 cores could be used like: One with less noise (low iso), other with high noise (high iso)and then “predict”

      • No, I can see where you might be able to do that. You’ve got two different signals, one of which is concentrated on part of the spectrum that often is problematic for noise. The IR isn’t filtered for anything but the red channel and above, then you’re getting some important information you can use. Indeed, just like doubled Green in a Bayer gives you extra resolution information, semi-doubled Red could give you extra noise information.

  • zuzu

    The new focus position control method described in the mentioned patent is completely different from both PDAF and CDAF. As far as focussing of a specific lens is concerned, it depends on two factors:
    1. Transmission of sufficient amount of IR light
    2. Focussing motor capable to move fast to prescribed position (like SWD)

    The second requirement is the same as for PDAF, but not CDAF.

    The bottom line: SWD lenses like 12-60 and 50-200 should focus very fast on those bodies, provided that they transmit enough IR light.
    The focussing speed of CDAF optimized lenses will increase only marginally.

    • As ‘my friend’ said “Amazingly fast with new lenses, slightly faster with older ones”. This could suggest an upgraded focus drive is installed. Those SWD lenses need a bigger piezo actuator (like in E3,E5).

    • As my friend said: Amazingly fast with new lenses, slightly faster with older ones.

      This could suggest an upgraded focus drive is installed. Those SWD lenses need a bigger piezo actuator (like in E3,E5).

      With SnR, the workflow maybe similar to E5 engine:

      • zuzu


        This piezo actuator is inside the lenses, not in the camera body. The actual limiting factor is the battery (those SWD drives require quite substantial current).

    • Walter

      That extra information is only interesting if it is correlated with the RGB information, though. The reason the clever Bayer demosaicing methods work is because of correlations between the colors (i.e. luminance channel has more high-frequency information).

      If the IR data is heavily correlated with the red channel (or even the luma channel) then you could get some extra information out of it. But if that’s the case, then how is the IR helpful for focusing? I don’t know the details about the new system, but I’m not sure how widening the spectral sensitivity of the sensor helps unless you’re doing some sort of bizarre active-IR focusing — and, if you’re doing that, then are you actually kicking out enough IR to make this system work at long distances?

  • Maley

    Why is the IR light needed? I dont understand what makes the af faster.

    • cbr09

      Because of Chromatic Abberations the IR image is in a slightly different place (behind I think) the visible. This means by comparing the two you know which way to focus. It is a bit like phase detect but does not require a special directional sensor. It means any point on the sensor can be used for focus – should be great for tracking.

      • Thanks for the explanation, that makes sense and sounds like a brilliant idea/patent. Hope it surfaces!

  • Mr. Reeee

    All this sensor and filter talk is pretty interesting, but what does it all mean in the real world, capturing images?

    • Olympius

      Darn you Mr. Reese, trying to inject reality into a conversation filled with theory and speculation! If you want reality you should go hang out at DPReview or some other place… :-)

      Seriously, we will know everything we need to know about this sensor and IQ in just a couple of days.

      Patience everyone, patience.

      But then, I have no plans on buying any of these Olympus offerings, so I could care less…


    • That – perhaps – mirrorless Olympus cameras also can be used for sports and action, in the future. Now that would be – as I see it – most definitely the next big thing and a big step on the way to realising Terada’s promise that Oly would merge FT and µFT “…into one beautiful system.”

    • Esa Tuunanen

      Trying to take my place as complaining cynic when there’s finally discussion of possible new break through tech instead of yet another unergonomic low end consumer camera?

      Plain contrast AF will always have nasty limitation of not instantly knowing how much and to what direction focus should be moved so if Olympus found a way to correct those weaknesses that’s great news.

  • cocute

    when Dxomark test new Olympus sensor and show better that old in RD and High ISO… We’ll believe

    • I don’t give a ***** about the doxmark results.
      In their world and with their testing approach, there never will be an FT sensor to deliver.
      I look the pictures my E-5 delivers and the WOW effect ist still there. Although I purchased the camera as early as last November.

    • This new AF technology has nothing to do with DR and high ISO. If anything, it would reduce the sensor area dedicated to capturing visible light (unless the IR channel can be used to evaluate the intensity of visible wavelengths, which is improbable).

      • cocute wasn’t talking about AF tech – there was much more to the admin OP than mere AF talk. I’m also more interested in real-world DR and low-ISO performance than AF. DXOMark will give a good idea, no matter what armchair scientists say.

        • I was trying to say that (if the IR technology is indeed implemented with this sensor) DXOMark will likely find that sensor noisier and lower in DR than the latest Panasonic sensors.

          • zune

            Possibly can DXOmark tell us but remember DXOmark use calculate only unlike visiblelight in sensor test all IR is out strainers.

    • Mr. Reeee

      DXO Test?
      Let’s just call the witch doctor and have him toss the bones. ;-)

      • and don’t forget to sacrifice a chicken or something…

  • Hi girls and guys.

    The great new is :Olympus is coming to make it own sensor…
    For me the great new is: Olympus instead of Panansonic follow the first great idea: a “12 megapixel sensor is enough.”

    It’s very important for who want to have better quality in low light iso en color depth… and for who do not want replace the computer every year…

    Thank’s oly, the EP3 with external EVF and spécial handgrip, become on time to put my Panasonic G1 in the cupboard

  • AirShaker

    Does it mean less back and forth pumping in video autofocus?

  • Hi Thom Hogan
    Can you explain why 12 mill pixels are not enough? I can print A3 size with my E1 and get perfect results.with low >iSO of 12 mill should be more than enough.especially for 43 sensor size…
    Thanks Shanti

    • croppers need as much mp as possible for instance..:-)

      • Nathan

        Or, you know, they could simply frame their shots before committing to the shutter release.

        If you intend to crop the crap out of your shots, 4/3 is entirely the wrong system for you. You need a 5D Mk II or D3x.

        4/3 is already cropped- just like 35mm, it’s small format photography, and even with 100 ASA film, you couldn’t get decent image quality from a crop comprising 20 percent of the film frame, and you can’t get it if your area is less than half the area of a four thirds cropped frame.

        Croppers should look elsewhere. Their habit of failing to compose a shot and use the right focal length might be an expensive one, but at least they don’t have to buy the right lens for their shot.

    • Rachnaroch

      He already gave you an example (Lytro), but there are many others, the most important being that people still demand more pixels. Your position is like saying a Toyota Camry is enough. Enough for who? For you? Them fine. But you are just one person out of billions. Myself? Yes, 12 MP are enough for me (and a Camry too). But I understand others may have different needs.

      On the other hand, I welcome improvements and for a given sensor size having more pixels IS an improvement if done right (you get more samples for a same surface). A 16 MP D7000 has better image quality than a 6 MP D70. In any aspect. How’s that bad? And you can always chose smaller image sizes if you don’t want all the pixels or don’t have enough computer power or storage. Or you can reduce the larger images to get away some of the noise or imperfections for even better IQ. Or print an A3 with more detail when you need it. More pixels give you more choices.

      • Strange enough, there are a lot of people – for instance at DPreview – who shoot with the 5MP E-1 as well as the latest Oly-technology and even state-of-the-art cameras of other brands who still claim that the E-1 delivers (at low ISOs, that is) the best IQ of all the cameras they had or have. So, MPs seems not to be as important as you claim it is.

        • Ditto the 4mp Nikon D2h. Older large photosite cameras have some advantages over current ones. For instance, the D2h had such large photosites that diffraction at f/16 didn’t really get recorded because it didn’t hit photosites far enough away from the original. Also the AA filters were tuned differently on low pixel count cameras because they didn’t resolve common problematic low-level patterns.

          Finally, a lot of people are getting fooled by the math. A 12mp sensor doesn’t “resolve” twice as much as a 6mp one. They resolve about 1.5x.

        • Yes the E1 still lives on…..the only reason for me to have more pixels is to get more detail out of landscapes,etc… but then I wonder is 12MP enough or does one need to use MF / 645 to capture enough detail?

      • Correct. More data is rarely bad news. In the Lytro case more data means that you can use computational ability to remove a lot of issues we have no solution for today, such as longitudinal spherical aberration. Then there’s 3D.

        There are a couple things you shouldn’t bet against in high tech (at least until the laws governing them appear to be coming to an end): computational power and bandwidth. As I wrote about as early as 2002, we’re going to see more and more that these things start playing a role in what we can do.

        • Rachnaroch

          Hey Tom, when will you start doing M4/3 guides? I know the market is tiny compared to Nikon’s but still…

          • Do you guys really want me to do them? I haven’t gotten much indication that the m4/3 crowd is clamoring for such things.

            • Chris

              It’d be a nice addition. It’d also help people who come across your site know that there are options outside of Nikon that aren’t Canon.

            • Nathan

              Uh, yes, Thom, we’re always interested in insight from experienced people. A micro four thirds camera has completely different strengths and weaknesses than a D7000 and as new gear comes out, it’s always interesting to read about what each piece of gear is good at or terrible at.

            • It’ll be nice if there would be more serious sites writing about the Mirrorless cameras,
              there are allot of confusion going around (thanx to pana/oly producing gazillion models) and people just dont know what to choose..actually you have written very nice “The Mirrorless Shootout”…
              if it will be constantly updated with newer cameras (and also the G3/GH2/NX11) i’m sure it’ll be popular and very helpful resource

    • Agent00soul

      Enough for what?

    • dumbo

      alot of pros are still shooting the 5d, not the mark 2 but the 12.8 megapixel original. it’s still good enough for magazines and bill posters, galleries etc, even in 2011 :)

  • I’m sill worrying about the “IR” pixels.

    Do these pixels help in increasing DR (and decreasing noise) or – as opposite – the captured IR information helps just in focus aquiring hence it’s simply dropped when processing the image?

  • ignoble.yeti

    “They told me it’s the best Micro Four Thirds sensor to date.”

    Seriously? Better than the GH2/G3 in terms of dynamic range and high ISO performance?

    • admin


      • Anonymous

        admin, you must be bursting beeing unable to reveal what you know in order to protect your sources…
        I am really looking into the m4/3 for my wife to catch my kids if it is good on C-AF. I probably stick to my E-330 and 50 F2 for dental photography – the limiting factor for size is the flash when shooting F22 and Olympus has the only RING-Flash (all others have ring-Diffusers with 2 strobes)

      • Anonymous

        admin, you must be bursting with info you cannot reveal to protect your sources… If C-AF is finally great I am lloking into m4/3 for my wife to catch the kids. Will keep myself my E-330 for dental pics as limiting factor is the flash when shooting F22 (and Olympus is the only true ringflash on the market opposed to ringdiffusers with two strobes from anyone else)

    • Yeah, that is a bit worrisome. We’ll have to see what Olympus says on Thursday.

      • Nah, surely far more fun to speculate one’s neck out Thom! lol

        Just read somewhere that perhaps Olympus are using the Panasonic 16MP sensor as the basis for this, and allowing for the IR sites providing a 12MP output image… could that make sense?

        Also, perhaps some use of the IR channel to reduce noise? and changes in filtering allowing more light through?

        Perhaps that could all add up to the improvement in DR and sensitivity over the GH2/G3?

        It sure will be interesting to see the dpreview analysis – I’m guessing very soon after announcement, given the interest in this sector.



        • Well, that would be more the Fujifilm approach: scatter focus elements through the sensor, but not 100%. But, no, that doesn’t exactly make sense to me. You get resolution gaps that way.

          • Dave

            Here’s an idea just to put out there…what if you take the higher MP sensor and overlay micro lenses in an alternating large/small pattern, with the larger lenses over the green sensels and the blue sensels, with the smaller lenses over the IR and red sensels. Could this reduce the gaps in resolution and help compensate for the loss of the extra green in the Bayer filter?

  • Steve

    I guess this explains the ‘R’ added to the end of the new kit lenses. Looks like they have just changed the lens coating to allow IR through to the sensor.

  • Dana Curtis Kincaid

    If the sensor is better they should put in in the E5 and call it the E5.2. There is precedence.

    • Steve

      You’ll have to wait another 3 years for that update. Actually… if this is true then the EVF is the last limiting factor for m43. This could easily be resolved in the next 3 years with Sony and Samsung now pushing the EVF advancement as well.

      • hmm..the latest aps-c sony sensor’s have 2 stop DR and 0.5 stop ISO advantage, frankly, olympus’s sensor might be closing the gap….but the upcoming Sony 24mpx sensor is supposed to be even more advanced…
        so its a cat and mouse game…while sony manages to be way ahead

        • Gianluca

          …but sony lens are far behind…so let’s see what will happen in the next years..;)

        • NEX will always be ahead simply because of much bigger sensor – unless Panasonic has some huge breakthrough in sensor tech. There’s much more to a camera system than a sensor, though.

          • Gianluca

            sony will always lead the mp race…but probably not quality and even more size race.
            The next sony sensor will be 24mp if you crop in m43 you’ll have a 15mp sensor circa.
            with same high iso and dr perfomance…..

            • Walter

    ’s attitude on NEX is basically “it would be nice, if you could find some lenses that didn’t suck to put on it.”

              • Nathan

                I had the same reaction when I went to look into the system at the local shops.

                Also, too few controls, menus are typical Sony- I.E. UGH.

  • Thomas S

    If some of the sensor area (25% according to the drawings in the patent) is dedicated to infra red, I am not sure if there will be a huge leap in DR and high iso performance. Of course, the drawings could be not to scale and the IR pixels could be smaller, but still…

    I hope that AF speed with FT lenses is decent with this new technology. This would be more important to me than over the top DR and high iso performance.

    • Not sure whether IR pixels do help in image rendering or not.

      I guess the IR component is simply dropped after focus aquiring but I also can imagine that it may help in luminosity channel reconstitution. In the latter case the DR should not deteriorate really much.

  • IR based AF prediction is an interesting development that might have been leak from manufacturing or applied researchers in Asia.

    First found here:

    I understand however that it works as a focus assist, simply predicting in which direction the focussing lens must move, while the fine collimation is still made by CDAF.

    Therefore it doesn’t disable previous lenses. One wonders if they might be recoated, or work up to a point.

    I moved the discussion to DPR because the site is faster (better than WordPress) , but there is a lot of competence here, so keep it coming.

    Admin I am still getting those ‘slow script’ messages, even if I have disabled moving ads in Apple Safari.

    • amalric,
      Look at Fig. 8 here:
      Apparently the system needs two measurements to evaluate the direction of the correction needed. However, it is predictive. Like PDAF, the difference between the images is indicative of the quantity of correction needed.

      I have no idea about the backward compatibility of the system regarding lenses. However, the redesign of the kit lens suggests that old lenses won’t work the same way.

      • This I think is the Colour Aberration principle at work to determine direction:

        IR might be there as a Yardstick, to determine how far the focussing lens must travel.

      • Thomas S

        Maybe the new new lenses have some extra abberation (hopefully only in the IR range) designed in?

      • Mal

        If the camera can calculate the amount of change in focus required, just as PDAF can, then would it not work better with PDAF lenses like the four thirds lenses?

        Now I am really excited…

  • SLOtographer

    Exciting news! The new sensor has been perhaps the main factor of this upgrade cycle in my mind. It seems to be driving AF and IQ, which are the two main things that have been identified as behind the competition.

    Outstanding questions that have been asked in the comments:

    1. How fast is the AF on 43rds lenses?
    2. How much did DR and high iso noise improve?
    3. Does it come with a coffee maker dongle attachment? Oh wait.

    • Anton Rahmadi

      (1) Slightly improved with older lenses (in microFT and 4/3)
      (2) can’t tell, but if it shares same sensor with GF3, then DR and SnR improvement are quite obvious.

  • greyhat

    My two cents:
    I think Thom Hogan is right and admin also.
    They may:
    1 – for better high ISO on files: have fixed blue channel issue and get better high ISO just on processor
    2 – for better high ISO on files: better manage sensor info to extract noise keeping (as possible) detail – kind of panasonics’ intelligent resolution
    3 – for better high ISO and DR on sensor output: lower sensor noise floor, change of circuitry (Thom Hogan’s option), this I think is less possible: requires IC design knowledge and quite embarassing to Panasonic (competition comes here and reworks circuit and lowers noise floor)
    4 – lighter bayer filters: if processor is better, color can still be recovered and allow more light to pass bayer filter, i.e. more light = better sensivity = better high ISO
    5 – tune sensor for a lower base ISO (ISO 100), this is not completely against 4 because photosites are bigger (12MP) and can be asked by Olympus to be made of deeper wells (can gather more light without saturation
    6 – use a different bayer array (CMYG, RGBWhite) to have better luminance and better high ISO; if you have better high ISO luminance overall IQ improve a lot (processor can soft only color and keep detail
    7 – get rid of anti-aliasing (or close to it) to increase detail at all ISOs, then processor can work on noise
    If you fix/tune/implement most or all these things you may be end up with a better IQ at all ISOs, i.e. a better sensor. It would not be a Pentax K5 (base ISO 80) level but it can be noticeable to user and more importantly to reviewers
    I _think_ sensor fab is Panasonic. Like Admin sources said gains on DR are on dark areas not brigh areas. Panasonic sensors are known for a very steep saturation curve with no histeresis (like sony and canon sensors), very different from film light sensitivity curve.

    • I wonder if they’ve modified the Bayer array from straight RGGB, whether it will affect RAW processing. Hopefully Olympus has contracted David Coffin to do the RAW support in dcraw, and eventually that spreads out to most of the other RAW processors, which either use a modified dcraw, or at least use it as a reference implementation. If not, it may be some time before all of the RAW processors are updated.

      IIRC, when Olympus went to the new format, starting with the E-410, it was something like 6 months before the complaints that Adobe, Lightroom, Bibble, RawTherapee, etc. did not support the new camera.

      • Ouch. Good point.

        Let’s hope Adobe/Apple/et al. are a little more ready to support a new format these days.

  • EyeAmLegend

    If Oly has a patent on this and this works well that could mean income from licensing or cross patent licesing for Olympus. More money = more development and cross licesing means better access to technology without having to pay for it.

    What I really like is how Olympus comes out with new ideas as the “underdog”

  • mahler

    I don’t know why it is good news that Olympus will be independant from Panasonic sensors.

    The only important thing is, if there are improvements visible in reality, and mainly in raw capture (the JPEG engine is software). In that sense, it does not matter who designs and produces the sensor.

    • Bobby

      If both O&P make sensors then there’s more incentive for each to outdo the other, and less chance that m43 will stagnate as a result of one company going through a bad patch. It therefore makes the format more stable.

      Also it means people are less likely to talk about ‘the m43 sensor’ (sic) in such a way as to imply that there are some physical laws that prevent m43-sized sensors producing excellent images at ISO 1600, for example.

      It looks to me like it could be an entirely new sensor. A lot of emphasis has lately been put on the word ‘tweaked’, but that’s a sufficiently unusual word to make me suspect it’s a single source. It’s entirely possible that the source may have just assumed that, as it’s a 12MP sensor and the pictures show an improvement, Olympus must just be getting better results from the same sensor. One reason the MP race still has some life in it is the fact that it’s the most obvious way of showing that the latest generation camera has a newer sensor in it.

  • Disraeli

    The sensor is RGGB,
    what happens is it some how shutters between RGGB and RGB-IR,
    with one of the G pixels, thats 1 in 4 exposed to IR light
    then it shutters back to RGGB

    sounds like a perforated IR cutoff filter over the sensor that is positioned over the photosites, and moves to shutter IR on/off for RGGB then RG-IR-B and back to RGGB

    this gives 2 references of light, one for IR the other for normal spectrum as RGGB. The two light indices have different focus values, as IR is slightly different focus to incident light. In this way the focussing is phased between to focus sources.

    Absolutely brilliant solution!

    • That would also be a mechanical solution, and thus prone to failure and manufacturing tolerances. I have to wonder if they’ve figured out a way to change filtration through electrical current stimulation.

      • Walter

        I’ve wondered about systems like that too. Is there any clever way to do that with liquid crystals? I know most of the tricks people do with them are related to polarization…

      • I haven’t read the patent, does it says that somehow it does both R-G-G-B and R-G-IR-B? Brilliant if it does (electrically, mechanically is simply unfeasible at that scale, unless you call “mechanics” things like eInk).
        Interesting if it doesn’t too, since the rumors goes that the images are better (how can they better if 25% of the surface is “gone” for AF?)

        • Disraeli

          No its nothing like a 25% loss, but nice try :)
          Autofocus mechanics are one thing, interpolating used AF portions are something else again (which is what the Fuji array does)

        • D

          I have read the patent. They use the difference in chromatic abberation of visible and IR light to determine focus.
          The R-G-IR-B is only used in the focussing area(s).
          The rest of the sensor is R-G-G-B. So no “switching” between IR/Color filters.
          Also, the patent states that the system could also work with only visible colors like red and blue. If that is true that is even better. Just design a lens with a lot of (predictable) CA to allow focus and correct the CA in software for the final image.

          • Thomas S

            Good point, I somehow missed it when reading the patent: R-G-B-IT is limited to AF sensor areas. With 35 AF points that is going to sum up, though. ;-) Just kidding. So loss of resolution shouldn’t be an issue!

            If the E-P3 can properly focus my FT glass. I’ll get one as soon as it becomes available. Somehow the mentioning of the interachangable grip gives me hope, and with the new IR AF you wouldn’t even need an adapter (integrated into the grip or not)!

            • Thomas S

              Of course you would need an MMF-1/2, but nothing fancy like an adapter with integrated mirror or sensor or the like.

          • Ok, so not ALL the sensor is AF-enabled but only some limited areas a la PDAF – makes sense.

            Regarding your idea, unfortunately you need “normal” CA for it to work, and that is not fixable in software. The CA we fix in post is lateral-CA which is not good for the focusing purpose:

          • That’s the Fujifilm approach. It also means that you’ve slight resolution deficiencies in your data.

      • Disraeli

        Prone to; wel perhaps yes, but IS has been 100% reliable and requires about equal tolerances. I am just guessing at how shuttering is achieved as I cannot think of another way to do it, not that it doesnt exist.

    • Neonart

      You down with OPP! Yeah you know me. BYOB. ASAP.

      (OPP=Other people’s patents, of course.)

    • Anton Rahmadi

      I guest you’re nearly there, I have to update my slide on :D

      • ding dong

        Am I the only one who misreads that website as “Mirrorless Scam” ?

  • Dana Curtis Kincaid

    Put a new sensor in the E5 and make it an E5.2!!!

  • Dana Curtis Kincaid

    What is the airspeed velocity of an unladen swallow?

  • Raist3d

    All very interesting. Well see how Olympus deal with possible resolution loss if they have the ideal going on. Reminds me of the qua photosite color Sony all in one bridge
    Camera- the 850 I think it was? With the emerald photosites.

    Sensor is not 16mp.

    Af will be really fast.

    • Disraeli

      There is no resolution loss to deal with. Interpolation would be exactly like the green pixel just next to it. All that needs to be done is the same green translucence be copied across.

      • Raist3d

        If you copy the same translucence and by that I mean by the adjacent green, that’s resolution loss. Maybe you mean something else? Or I misread what you wrote.

  • Raist3d

    All very interesting. Well see how Olympus deal with possible resolution loss if they have the IRl going on. Reminds me of the quad photosite color Sony all in one bridge
    Camera- the 850 I think it was? With the emerald photosites.

    Really doubt sensor is not 12 mp

    Af will be really fast.

  • Oliver

    The Industrial Solutions division of Olympus uses a custom sensor for this camera:
    Look at the specifications, it says custom CMOS sensor. Maybe Olympus has the capability of designing it’s own sensors.

    • Thomas S

      It also has a global shutter. Come on Oly, let all that cool stuff trickle down into the Pens! ;-)

  • Thomas S

    Why not combine Fuji’s on-sensor PDAF with Oly’s IR-AF? ;-)

    Not sure if this would speed up things even more, but definitely worth a try. Maybe Fuji and Oly can cross-licence these technologies to each other? Wait a minute: wasn’t there a rumour about a Fuji MFT camera? If that ever materializes I’m curious to see what AF tech they’d use!

    In all these experiments, however, care should be taken not to obfuscate too much of the sensor: lightning-fast AF, but pictures are all black ;-)

  • It will be funny if it is a nicely tweaked sensor, and actually is better in many ways, and yet still gets a score of 56 or so by DXO labs. then you will see marathon threads about how Olympus has failed again. I’m more interested in the new lenses and the overall camera improvements, unless the sensor changes are major.

  • Steve-O

    ignoble.yeti8 hours ago | Reply: “They told me it’s the best Micro Four Thirds sensor to date.”

    “Seriously? Better than the GH2/G3 in terms of dynamic range and high ISO performance?”

    admin8 hours ago | Reply: “Yeah!”

    Happy dreams tonight, and may stay up until 1 AM tomorrow night (6 AM in London) just to check out the announcement. Whether new design or redesign, sounds like break-through performance improvements in DR, high ISO and AF speed, enough for me!

    I’m also very curious about the VF-3. So far all I’ve heard is smaller and cheaper than the VF-2, nothing about performance.

    Admin, any fuzzy photos of this?

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.