u43 sensor, dynamic range improvement and aa filter ??

Discussion in 'This or That? (MFT only)' started by LovinTheEP2, Jul 29, 2011.

  1. LovinTheEP2

    LovinTheEP2 Mu-43 Top Veteran

    619
    Feb 15, 2011
    Toronto
    Just curious if anyone knows the answer to the following questions I have been thinking about lately.

    For a significant improvement in the u4/3 sensor, is it specificallly a sensor design issue/limitation or can a processor developed by Olympus or Panasonic make a DR improvement by 2-3 steps to bridge it to a NEX level. I don't believe it's the size of the sensor (slightly smaller size then the NEX) that is handicapping it. Just curious if it's something Olympus might be able to do independantly or if they are tied to an improved sensor design by the manufacturer to make any gains in DR.

    Also, the Fovean sensor in SD1 doesn't require an anti-alias filter to control moire.. does anyone know such a feat could be done with the u43s platform in the future.

    If the u43 sensor had a significant improvement in DR and a super weak or lack of aa filter to improvement ooc raw/jpg sharpness, it would easily be the system to have now that the slower focusing aspect has been addressed. If they'd do such a thing in the PRO model, I'd put my deposit and preorder down immediately.
     
  2. Ned

    Ned Mu-43 Legend

    Jul 18, 2010
    Alberta, Canada
    Olympus does exactly that, and they started it in a pro body (the E-5). The E-PL2 and every generation of Olympus body since features an almost clear AA filter, which is possible by filtering out of moire, pseudo color, etc. through the Trupic V+ Image Processor featuring "Fine Detail Processing".

    Moire patterns are far better controlled using the Olympus system than with the Foveon system (which still has a problem with this), but Olympus uses the next weakest AA filter after Sigma.

    Your E-P2 unfortunately doesn't have the weak AA filter. The E-PL1 has a weaker one than the E-P2, but the E-5 (E-System DSLR) and E-PL2 have the weakest AA filter for a non-Foveon sensor, as well as the next generation on (the E-P3, E-PL3, and E-PM1).
     
    • Like Like x 2
  3. WT21

    WT21 Mu-43 Hall of Famer

    Feb 19, 2010
    Boston
    Here's what I understand. Perhaps there are others who know better:

    I believe the newer Olys have a weaker AA filter, which is what improved their sharpness over the EP1.

    Output can be improved possibly in jpg, but physical sensor size really does impact IQ. Lenses have to be built much better, and sensels (I think that's the turn) "fill up" (i.e. overexpose) with light more easily when smaller.

    Oly is handicapped by the fact that there is only one 43 sensor maker currently -- and that's Panny. And Panny won't sell the better 43 sensors to Oly, because it's their competitive advantage. That's why people get excited about the talk of Oly partnering with Fuji and others who make their own chips -- to get in another chip maker to the mix, and free up the best sensor from being reserved for only Panny's top of the line camera.
     
  4. LovinTheEP2

    LovinTheEP2 Mu-43 Top Veteran

    619
    Feb 15, 2011
    Toronto

    Thanks for the feedback. I didn't know Olympus had such a weak aa system. Guess that might explain why I prefer the Olympus files and output moreso then Panasonic.

    Any idea on if Olympus can do anything about improving DR of the u43 sensor of it's a design limitation???
     
  5. kahren

    kahren Mu-43 Regular

    145
    Mar 21, 2010
    nex sensor has an area which is 40% larger then that of the m4/3 sensor (gh2 is bigger), thats not "slightly smaller"
     
  6. Ned

    Ned Mu-43 Legend

    Jul 18, 2010
    Alberta, Canada
    This has already improved since removing (mostly) the AA filter, as you had surmised. It's not going to be a perfect system in every way, though. I don't know why people are so hung up on DR unless Landscape Photography is all they do. Even if Landscape Photography is your thing, you still have to weigh in the advantages of the Four-Thirds or Micro Four-Thirds system against the disadvantage. Does broad DOF without loss of lens speed, no vignetting, edge-to-edge sharpness, edge-to-edge brightness with even exposure, and lessened Chromatic Aberration count for more than the slightly limited Dynamic Range?
    Then there's other factors that could weigh in depending on your needs, such as (some of these only relate to Olympus DSLRs, not PENs) In-Body Image Stabilization, the only Dust Reduction System which actually works (along with Panasonic of course), 2x crop factor for telephoto lenses (and extra short focal lengths to compensate for the wide angle end), the best JPG engine for great colors, lowest JPG compression, Remote Commander in every body that features a built-in flash, the toughest weather-sealed build and system (along with Pentax), the brightest Optical Viewfinder (in pro-grade DSLRs), and the Super Control Panel menu system.

    So will Olympus ever have as good Dynamic Range as Full Frame systems? At the rate they're going they'll soon catch up to where Full Frame is now, but by then Full Frame may or may not have improved. It may fall behind, but will always get better to meet most people's needs. The question is if the other systems will ever catch up to Olympus on their myriad advantages? At the moment, it doesn't seem like they're making any progress at all in those areas, and it doesn't look like they care enough to try. At least Olympus has the pressure to improve on what the "big boys" are considered good at, and have caught up and even surpassed in many ways in the areas that they have been criticized for when compared with the big two. For instance, PEN Autofocus Speed is now better than Canon and Nikon DSLR AF speed. Who would have ever thought? High ISO sensitivity is almost there. The system is growing and improving faster than any other that I can tell.
     
  7. WT21

    WT21 Mu-43 Hall of Famer

    Feb 19, 2010
    Boston
    Also if you are hooked on landscape, you should/could use things like graded ND filters, CP filters, etc.
     
  8. carpandean

    carpandean Mu-43 Top Veteran

    827
    Oct 29, 2010
    Western NY
    Actually, to get technical, the :43: sensor is about 40% smaller than the NEX sensor. That is, if you take away 40% of the NEX sensor, you get the same area as the :43: sensor. This means that the NEX sensor is actually 64% larger than :43:. You have to add another 64% of the :43: to it to get the same area as the NEX sensor. The Canon APS-C is closer :)43: is 31% smaller or APS-C is 46% larger.)

    It should be noted, though, that :43: is closer to APS-C than APS-C is to full frame. APS-C is 58-62% (depending on which brand) smaller than FF, meaning that FF is 133-163% bigger than APS-C.
     
  9. starlabs

    starlabs Mu-43 Top Veteran

    856
    Sep 30, 2010
    Los Angeles
    Also, you can always use HDR - which while it's still more work - also pretty much IMO negates any issues with "low" DR of the m43 sensors.
     
  10. LovinTheEP2

    LovinTheEP2 Mu-43 Top Veteran

    619
    Feb 15, 2011
    Toronto
    How is DR solely based on the size of a sensor???

    Maybe that is were I guess I didn't understand sensor dynamics. I know ISO performance (which Olympus should be able to control on it's own chipsets) affects DR but why does a larger sensor mean better dynamic range???

    A larger sensor captures more light obviously but I would imagine it also to do with the material and layout of the array's that affect the quality of the light gathering and how that data is analyzed. Hence why i was wondering if Olympus on it's own could improve the dynmaic range of the sensor simply by better coding from it`s chipset (like what it has done with AF).

    I don`t believe a high DR sensor is necessarily only required for Landscape. In that case, there is a number of options available to you regarding lens filters or stacking of images to deal with improvement of noise and getting more dynamic range out photo.

    Dynamic range is also useful in my opinion for portrait work and general action photography as well as its provides a smoothness to color transitions and highlighting local micro contrast of sharp color transition (face, clothing etc.). That is one area where I feel the current m43s system is a bit weak compared to the NEX. That from a single out of camera exposure without addon filters which would all affect the exposure speed by reducing light.

    Granted, as was nicely pointed above... the m43s system advantages in my opinion outweight the better Sony sensor and even the nice X100 accurate auto wb and color gamut.

    Sigmas lack of evolving the Fovean DP1-DP2 systems other very minor tweaks made me convert to and purchase a u43 camera even though I still like the output from a DP1s better then from my EP2 but the DPs have way toooo many drawbacks outside of imagine quality once the red issue was resolved.

    Its nice to see Olympus continuing to improve the PEN system (albiet I still think they have a marketing problem and tooo many different cams with little purpose between them to justify having 3 cams on the go... a PRO in the mix and seriously..) and justify the purchase of lens as it seems the system is not going to be dropped or obsoleted by a competitor anytime soon. Just curious if a significant improve in DR is at all in the pipeline or if its simply a limitation of the format size and any real future gains are solely in ISO performance (maybe another stop) and whether a future model has a built in vf (which is a nonfactor for me) and marketing addon tweaks to keep the line refreshed.

    Reason: EP3 is a great step forward from the EP2-EPL2 but maybe ... not enough of one in regards to photo quality of less then a 1-stop improvement in ISO unless the fast autofocus is a major issue for the upgrade and the onboard flash (which for me might be a reason alone to upgrade). Hence why im on the fence about a transition to the EP3 or just go with a Mini.
     
  11. ~tc~

    ~tc~ Mu-43 Hall of Famer

    Oct 22, 2010
    Houston, TX
    It really comes down to pixel density, so generally, larger sensors have an inherent advantage, but only so long as the MP are about the same. If Sony comes out with the rumored 24 MP chip in APS-C, in theory, it would be a close contest on DR assuming the same generation of technology between the two chips.

    Currently, though, the NEX sensor is "more better" than the size alone would suggest, meaning they are taking better advantage of current chipmaking technology than m43 are, and one would reasonably expect that to remain true with the 24MP sensor.

    I guess to sum up, the OLD sensor is more of a problem than the SMALL sensor.

    FYI, our "crappy" sensor still has far more DR than any color film ever made (and not far behind any B&W film ever made), and that was "good enough" for most of the greatest images ever recorded.
     
  12. RobWatson

    RobWatson Mu-43 Hall of Famer

    It is not. A general rule is that smaller pixel holds fewer electrons and thus has less DR. For high DR one needs to collect many photons and convert them into many electrons. As with all generalities there are variations but typically not outright exceptions. Our friends the sensor designers and manufacturers are constantly finding new ways to stuff more electrons into smaller pixels. Certainly one can find several different sensors with the same pixel size (in terms of square microns) but they can have very different capacity and thus DR.

    Tiny sensors with exceptionally large pixels can have tremendous DR due to the larger capacity to contain electrons. One of my favorites is the E2V CCD47-10 series of ccds that has 100,000 electron capacity and output bandwidth properties capable of ~3 electrons read noise and DR of 25,000:1 are not out of reach. The pixels are 13x13 microns and there are 1024x1024 pixels. One can almost get away with 1" format c-mount optics as the sensor is only 18.8mm diagonal.
     
  13. Hikari

    Hikari Mu-43 All-Pro

    Nov 26, 2010
    Both color and B&W negative film have more dynamic range than an m4/3 sensor.
     
  14. Pelao

    Pelao Mu-43 Top Veteran

    959
    Feb 3, 2010
    Ontario, Canada
    How do you know that Panasonic won't let Olympus use the latest sensors?

    I have read similar comments in a number of places, and wonder is it just assumption, or is there some evidence?
     
  15. WT21

    WT21 Mu-43 Hall of Famer

    Feb 19, 2010
    Boston
    Fair point. No idea really, but the fact is Oly has the old Panny sensors. Maybe Oly just don't want to pay for them, or maybe they wouldn't fit Olys IBIS infrastructure. In this case, I am just parrotting what I heard on the internets.
     
  16. Amin Sabet

    Amin Sabet Administrator

    Apr 10, 2009
    Boston, MA (USA)
    Neither Pana nor Olympus have explicitly stated this, but a Pana exec did point out - in the context of discussing why the GH1 sensor didn't make it into an Oly camera - that the two companies are competitors. That doesn't necessarily mean that Pana wouldn't give them the sensor. It could, for example, mean that they weren't inclined to give Oly a terrific deal on the sensor :rolleyes:.

    As it turns out, sensor size doesn't explain the degree of disparity in DR between current 4/3 sensors and the DR class leading APS-C sensors. For example, if you put the NEX-C3 sensor in a 4/3 camera and simply used the central 17.3 x 13mm of that sensor discarding the rest, you'd have a 4/3 sensor with nearly as much DR as the whole NEX-C3 sensor. In terms of stops, sensor size has an exponentially greater effect on signal/noise than it has on DR.

    The main issue limiting DR with our sensors is that Pana lags behind Sony in terms of designing sensors with very low read noise. Canon likewise lags behind too. Canon used to be king in sensor design, and now it's Sony. No doubt things will change again at some point, hopefully in favor of those designing sensors for our system :wink:.

    As far as AA filters go, the E-P1 and E-P2 had relatively stronger AA filters, which made them the least susceptible to aliasing but slightly less effective in capturing pixel level detail. All of the Panasonic cameras use weak AA filters, and the E-PL1/2 and E-P3 do as well. The E-PL1/2 and E-P3 offer more detailed JPEGs than the Panas, but that is a function of the JPEG engine. On a RAW level, all the Panas hang tough with the E-PL1/2 and E-P3 in the detail department. I've tested each and every one of them.

    In terms of real photography at normal display sizes, I find all of this dynamic range business to be pretty unimportant and personally did not sweat for an instant going from the GH2 to the E-P3, which is a step backwards in DR at base ISO. The difference in AA filter strength is probably even less important, although I confess to getting a warm fuzzy feeling when I zoom to 100% in Lightroom and see crisp detail with minimal sharpening needed.
     
  17. RobWatson

    RobWatson Mu-43 Hall of Famer

    ? In my book Dynamic Range = DR = Signal /noise = SNR ... clearly different meanings are intended.

    I sure would like to hear why sensor size is considered to have the effect stated.
     
  18. Pelao

    Pelao Mu-43 Top Veteran

    959
    Feb 3, 2010
    Ontario, Canada
    Regarding sensor price, you could of course be right. I find the case a bit suspect though. Olympus left their future very vulnerable if they committed to a format without a reasonable agreement regarding future sensors. It doesn't seem to make much sense. All speculation, but I bet it has more to do with overall margins on cameras, philosophy regarding what will sell and overall model range.

    "In terms of real photography at normal display sizes, I find all of this dynamic range business to be pretty unimportant and personally did not sweat for an instant going from the GH2 to the E-P3, which is a step backwards in DR at base ISO. "

    I know this has been discussed before (mainly around your post a few months back), but I feel the devil is in the detail of your statement. In other words, it spends upon what you consider to be normal display sizes, and it depends upon what is real photography for the individual shooter. Generally, I agree with your point, but when shooting landscapes I see a difference between M4/3 and my Canon 5D. It's there, it's real and for my aesthetic and quality sense, it is relevant, and requires a conscious decision and care around technique, especially at low ISO.

    It's a compromise I am willing to live with, and happily. DR may or may not be relevant to an individual's photographic output. I am willing to bet that when comparing M4/3 to other sensors, better quality output is more likely to be influenced by skill, lens, PP and subject than by differences in DR.
     
  19. Amin Sabet

    Amin Sabet Administrator

    Apr 10, 2009
    Boston, MA (USA)
    I definitely understand that the differences can be important to some people. When I say that "I find" it "pretty unimportant", it's really a personal statement, although I suspect there are plenty of others who share my lack of sensitivity to these issues and yet are convinced by reviews that the differences should be significant to them.
     
  20. LovinTheEP2

    LovinTheEP2 Mu-43 Top Veteran

    619
    Feb 15, 2011
    Toronto
    From the above few posts, it does seem the DR limitation currently inheriant in the Pana manufactured u43 sensor is the rate limiting factor for improved DR potential and not a processing engine issue for which, Olympus could potentially improve upon with better coding and or chipsets. Good to see others agree that its more of a Panasonic issue then format-size issue that is holding an improvement in DR.

    That being the case, there isn't a reason to believe that a significant improvement is in the cards in the foreseeable future unless it's a special edition sensor for Pro based models is announced, if those ever come to be since the evolution from the EP-1 to EP-3 are more to do with weakening the aa filter and improvement in focus abilities then improving the sensor directly.

    Curious to see what Fugi or Leica do in the next bit but those wont come to market for probably at least a year .. if not longer.

    Seems like a holding pattern for the next 12 months... to bad the mini isn`t out yet so i could make up my mind which whay to go.