A future technology for dSLRs?

Discussion in 'Digital Cameras' started by Ray Fischer, Nov 21, 2009.

  1. Ray Fischer

    Ray Fischer Guest

    Given that SLR makers are already incorporating piezo dust shakers,
    it might not be too far a stretch to include this technology as well.
    It would deal with the Bayer sensor's shortcoming and probably end the
    use of the Foveon sensor.

    Hasselblad has announced a Multi-Shot (MS) version of its H3DII-50
    medium format camera. First shown in the H3DII-39 MS in 2008, the
    system captures four shots in a row, moving the sensor by one
    pixel between each shot to record full RGB values at each
    position. THe H3DII-50 MS costs 23,000 euros with less expensive
    trade-in prices and a trade-up route for Hasselblad owners.
    http://www.dpreview.com/news/0911/09111902hasselbladh3d50ms.asp
     
    Ray Fischer, Nov 21, 2009
    #1
    1. Advertisements

  2. Ray Fischer

    Charles Guest

    Interesting idea. I looked closely at the armor shot and could see more
    detail. But, it was not a drop-your-socks event.

    How does one keep the required image shifts to exactly one pixel? In two
    dimensions? Over a period of time? After the camera has been thumped once
    or twice? The cost disadvantage is obvious (at least at this point in
    time).

    I'm still rooting for Foveon II, as it seems more realistic to me as a way
    to leave (progress beyond) Bayer demosaicing.

    Thanks for the post.
     
    Charles, Nov 21, 2009
    #2
    1. Advertisements

  3. Ray Fischer

    Guest Guest

    don't hold your breath on that one, and bayer actually works quite well.
     
    Guest, Nov 21, 2009
    #3
  4. Ray Fischer

    Charles Guest

    Agreed as to both points. However, the Foveon technology intrigues me with
    its elegance ... just like nuclear fusion. Probably both great ideas that
    will never actually pan out at any practical level.
     
    Charles, Nov 21, 2009
    #4
  5. Ray Fischer

    Guest Guest

    foveon is anything but elegant. it doesn't actually sense pure red,
    green and blue like in the ads. there's a lot of overlap in the three
    layers, requiring a *lot* of processing to produce r/g/b, and because
    of that, the results are less accurate than bayer with a lot more
    noise, even in ideal conditions. it's tough to break the laws of
    physics.
     
    Guest, Nov 21, 2009
    #5
  6. Ray Fischer

    Charles Guest

    I was not clear enough: the "idea" is elegant. Future vertical processing
    of semiconductors offers hope in this arena. There is a slim, but real,
    possibility that defined energy gaps could be built into a vertical
    photosensor.

    Planar to vertical and planar plus vertical might just give us some
    wonderful new sensor technology. Or not.
     
    Charles, Nov 21, 2009
    #6
  7. Ray Fischer

    Ray Fischer Guest

    Dunno. The fallback is to do the usual Bayer processing. Another
    option is to spit out an error.

    Sure, it's a feature with limited need, but I can see it being a cheap
    add-on to an existing camera.
     
    Ray Fischer, Nov 21, 2009
    #7
  8. Ray Fischer

    Guest Guest

    the idea is interesting, but it isn't really solving a problem that
    needs to be solved. the eye can't see the added colour resolution.
    nikon has a dichroic mirror patent but it's going to be a bitch to
    manufacture. fuji was working on organic dyes but i think they gave up
    on it. canon had a controllable filter, but it would require multiple
    exposures. and nobody mentions that the data processing and storage
    needs go up threefold (at least) with one of these systems.

    meanwhile, bayer works.
     
    Guest, Nov 21, 2009
    #8
  9. Ray Fischer

    Guest Guest

    i assume by zoom you mean pixel peep, and if you do that you can see
    all sorts of problems. why not look at photos normally? did you grain
    peep film too?

    the eye's colour resolution is about 1/10th its luminance resolution
    and bayer is 1/2. in other words bayer already has more colour than the
    eye can resolve, so there's not much point in making it even higher.
     
    Guest, Nov 22, 2009
    #9
  10. Ray Fischer

    Guest Guest

    it's not that people hate it, it's that there's a lot of drawbacks that
    some people blindly dismiss. a full colour sensor would be better *if*
    it didn't compromise on high iso, dynamic range, frame rate and colour
    accuracy. it also triples the storage and processing requirements.
    foveon loses on all of them. nothing is perfect, and bayer is an
    excellent tradeoff of what is currently possible.
    they're moving the sensor and combining the results. that's not the
    same as a full colour sensor and it certainly won't work very well for
    anything with motion in it. it's also not cheap.
     
    Guest, Nov 22, 2009
    #10
  11. Well, colour still photography and colour film by shooting through
    3 different filters is actually a very very old technology.
    It never worked well with moving subjects, why should it work
    better for Hasselblad? Do they change the laws of physics?

    As for why H is doing it ... because people like you decide that
    photographers have to have full colour sensors or you won't use,
    buy, etc. their shots. So everyone must buy H.

    -Wolfgang
     
    Wolfgang Weisselberg, Nov 22, 2009
    #11
  12. About as much overlap in the filters on a Bayer sensor.

    Go educate yourself, troll.
     
    Outing Trolls is FUN!, Nov 22, 2009
    #12
  13. Ray Fischer

    Better Info Guest

    If the sensor is shifted in sub-pixel increments, then much higher
    resolutions are available for both chroma and luma.

    http://www.photoacute.com/mtf.html
     
    Better Info, Nov 22, 2009
    #13
  14. Ray Fischer

    Guest Guest

    significantly more. see the foveon patents.
     
    Guest, Nov 22, 2009
    #14
  15. Ray Fischer

    Guest Guest

    since bayer cameras have as much as 5 times as many pixels (in the same
    35mm form factor), you can make substantially bigger enlargements than
    with foveon. if you step up to medium format backs, it's over 12 times
    as many pixels.

    do you regularly make 20" x 30" posters?
     
    Guest, Nov 22, 2009
    #15
  16. Ray Fischer

    Guest Guest

    of course it would be preferable *if* you could do it without
    compromising anything else. unfortunately, there's no free lunch.

    the reality is you end up compromising on things which are far more
    important than having higher colour resolution. you *will* see higher
    noise and colour casts, but you won't see the added colour resolution
    and a slow frame rate is not desirable for some types of photography,
    such as sports.
     
    Guest, Nov 22, 2009
    #16
  17. Ray Fischer

    Guest Guest

    completely wrong.
    yes, wild guesstimates that are completely bogus.
     
    Guest, Nov 22, 2009
    #17
  18. why not go the whole nine yards, use a monochrome sensor (no bayer
    pattern at all) and a slew of filters? In that way you'd not
    only have some broad filters which sort-of center on red, green
    and blue --- you could use narrow spectrum filters in addition
    and IR-pass and UV-pass filters and cyan, yellow, magenta and
    emerald and filters for all the difficult to resolve colours, too.

    Works perfectly well in our remote probes, in space, currently
    on Mars, too --- even worked on the hot and inhospitable Venus.

    It doesn't need a sensor shift mechanism but uses well known and
    time-tested methods. It also offers much higher effective ISO
    values for monochrome shots when used with a clear or missing
    filter, as no Bayer filter reduces the passing light.

    But if you insist on a sensor shift, make it shift sub-pixel
    distances ... 16 (i.e. 4x4) images could quadruple the resolution.
    The software to do so is already available, look e.g. here
    http://auricle.dyndns.org/ALE/gallery-auto/

    In addition, you can use it for many scientific tasks by just
    choosing the correct set of filters.
    You get close to quadrupling as often as you win the national
    lottery. It *could* happen, theoretically ...
    The average case would be about 30-50% more resolution.

    Remember that the green filter is broad, and records also red
    and blue light, just weaker.

    -Wolfgang
     
    Wolfgang Weisselberg, Nov 22, 2009
    #18
  19. Ray Fischer

    Guest Guest

    all of sigma's current cameras (including the vaporware sd15) have the
    same 4.7 megapixel sensor they introduced a few years ago. meanwhile,
    nikon, canon and sony have sensors as high as 24 megapixels. dividing,
    24 by 4.7 is 5.11 times as many pixels.

    phase one has a 60 megapixel medium format back. 60/4.7 = 12.77 times
    as many pixels.
     
    Guest, Nov 22, 2009
    #19
  20. Ray Fischer

    Guest Guest

    that's hardly a free lunch. it's only free if you aren't moving, which
    rules out just about everything other than still life with the camera
    on a tripod.
    and that 2/3 is regenerated with excellent accuracy. that's why it's so
    clever.
     
    Guest, Nov 22, 2009
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.