230px-ccd-sensor.jpg

The Future of the Digital Camera Sensor – TWIP

I’ve had some rare free time to think this week – just think. And in the course of firing up the old synapses, I decided that I have an idea of what the digital camera sensor will look like in five years or so…

I make these predictions having SOME inside information. I have spent some time with engineers at one of the major camera manufacturers in Japan. I received some insights there. But more than half what I write here is conjecture.

This list is just a starting point for me. Nobody can say I am right or wrong, because nobody knows. You can, alongside my conjecture, come up with some of your own, and add it to the comments section.

GOING FORWARD

a. The sensors of the future will al be full-frame. The technology will exist to make these things for pennies, removing the need for cropped sensors.

b. Price – well that’s easy. It will go down – way down. Future sensors will be far more advanced and cost far less to produce.
c. New sensors will be capable of very high dynamic range. In the old days when we shot B&W film, it had about 11 stops of dynamic range (or the ratio of maximum light intensity measurable, to minimum light intensity measurable.)

The digital sensor of the future will exceed that and closely mirror the human eye.

d. Light sensitivity will be improved. ISOs of 12,800 will look as good as ISO 250 today.

e. Future digital cameras will capture 64 bits of data.

f. The future sensors will all incorporate some sort of vibration reduction, eliminating the need for stabilized lenses.

g. Megapixel madness will have finally stopped at about 100 megapixels with most cameras offering sensors in the 25-50 megapixels range.

h. The new sensors in digital cameras will be more like high-end digital movie cameras, incorporating three chips rather than one. Each chip shall be responsible for red, green and blue color capture rather than the current all-in-one RGB capture.

i. Future digital camera sensors will come with special coatings making them 100% dust resistant, eliminating the need for sensor cleaning.

j. Image processing algorithms will be significantly improved, potentially eliminating the need for separate image processing engines. Higher-end chips will do DSP on-board.

Join the conversation! 63 Comments

  1. 25-50 MP at 64-bits each…. :( Guess I’ll start saving for four 10 TB hard drives and a Drobo Firewire 3200.

  2. 25-50 MP at 64-bits each…. :( Guess I’ll start saving for four 10 TB hard drives and a Drobo Firewire 3200.

  3. Re: points (c) and (e)

    one stop doubles the light, correct?
    So 11 stops is 11 bits of dynamic range.
    Wouldn’t 64-bit sensors provide 64 stops of dynamic range?

  4. Let’s hope storage will evolve as fast as digital sensors and processors.

  5. Let’s hope storage will evolve as fast as digital sensors and processors.

  6. Scott –

    On item a, the full frame sensor, I think you’re on the right track, but I don’t think we’ll completely migrate to full frame sensors.

    I’m guessing that the the current APS sensors wil co-exist with the full frame just as 35mm co-existsed with medium format.

    It may end up being a niche, but I think there will be a place for a smaller sensor that enables the design of smaller, lighter lenses.

    GS

  7. Scott –

    On item a, the full frame sensor, I think you’re on the right track, but I don’t think we’ll completely migrate to full frame sensors.

    I’m guessing that the the current APS sensors wil co-exist with the full frame just as 35mm co-existsed with medium format.

    It may end up being a niche, but I think there will be a place for a smaller sensor that enables the design of smaller, lighter lenses.

    GS

  8. You are missing GPS tagging. GPSes are being pushed into smaller and smaller devises. Next would be wireless. Try to think of the camera as another peripheral just like the keyboard mouse printer/scanner, etc. Possibly replacing the glass in the lens with something; it may just be that episode of Star Trek when they go back in time to bring back the whales and they developed the plastic to replace the glass but that idea has always stuck with me. Next would be a specialty camera or single purpose cameras. Finally I think you are being too high minded. Think Gimmicks and lots of them (i.e. pop up umbrellas, pink cameras, and new marketing names like the sniper or Porsche branded Sonys)

  9. You are missing GPS tagging. GPSes are being pushed into smaller and smaller devises. Next would be wireless. Try to think of the camera as another peripheral just like the keyboard mouse printer/scanner, etc. Possibly replacing the glass in the lens with something; it may just be that episode of Star Trek when they go back in time to bring back the whales and they developed the plastic to replace the glass but that idea has always stuck with me. Next would be a specialty camera or single purpose cameras. Finally I think you are being too high minded. Think Gimmicks and lots of them (i.e. pop up umbrellas, pink cameras, and new marketing names like the sniper or Porsche branded Sonys)

  10. I’ve believed that for many years that not only will dynamic range and light sensitively be improved, but the entire way a light sensor controls that sensitivity will be changed in a manner that allows the ISO of each individual pixel to be fine tuned in real time. This will allow the camera to deal with lighting changes in much the same way our own eyes do.

  11. I’ve believed that for many years that not only will dynamic range and light sensitively be improved, but the entire way a light sensor controls that sensitivity will be changed in a manner that allows the ISO of each individual pixel to be fine tuned in real time. This will allow the camera to deal with lighting changes in much the same way our own eyes do.

  12. A poster wrote: “25-50 MP at 64-bits each…. Guess I’ll start saving for four 10 TB hard drives and a Drobo Firewire 3200.”

    Yes, but this will cost about $79.95 by that time! :-)

  13. A poster wrote: “25-50 MP at 64-bits each…. Guess I’ll start saving for four 10 TB hard drives and a Drobo Firewire 3200.”

    Yes, but this will cost about $79.95 by that time! :-)

  14. Don’t the Foveon X3 chips already do h? Well I guess it is not three chips for each color but it does get reach color in their own space.

  15. Don’t the Foveon X3 chips already do h? Well I guess it is not three chips for each color but it does get reach color in their own space.

  16. What about the size, weight and cost of the lenses to cover with decent quality levels those “full frame” sensors?
    By the way, for example, Four Thirds is full frame… for Four Thirds.
    24 x 36 mm is a relic from analog times. I could believe for example in a 24 x 32 mm sensor.
    I am with George Ginos…

  17. What about the size, weight and cost of the lenses to cover with decent quality levels those “full frame” sensors?
    By the way, for example, Four Thirds is full frame… for Four Thirds.
    24 x 36 mm is a relic from analog times. I could believe for example in a 24 x 32 mm sensor.
    I am with George Ginos…

  18. I’m curious about your multi-sensor speculation. While I’ve never looked at the mechanical drawings of today’s high-end video equipment, I assume there’s some sort of prism that splits the image for the three high-quality sensors. This would require three times the physical space, correct? That can’t bode well for the body sizes, especially if you’re correct about full-frame sensors.

    I think there will continue to be a market for crop sensor bodies though – the “free” extra focal length you get as a result is great when you need a long lens. As a bird photographer I’m sure you’ve always felt the need for having more reach, and a 500mm lens on a 1.6x body yields the same effective focal length as the ludicrously expensive (and currently unreleased as far as I know) 800mm super-tele on a full-frame body. I’d love to have a FF body for the wide-angle shots without needing specialty EF-S lenses and of course the additional image quality, but it’s the really long lenses that cost the big bucks and at this point in my career it’s not worth the extra money for the image quality gained (discounting the additional features only available on FF bodies, of course).

  19. I’m curious about your multi-sensor speculation. While I’ve never looked at the mechanical drawings of today’s high-end video equipment, I assume there’s some sort of prism that splits the image for the three high-quality sensors. This would require three times the physical space, correct? That can’t bode well for the body sizes, especially if you’re correct about full-frame sensors.

    I think there will continue to be a market for crop sensor bodies though – the “free” extra focal length you get as a result is great when you need a long lens. As a bird photographer I’m sure you’ve always felt the need for having more reach, and a 500mm lens on a 1.6x body yields the same effective focal length as the ludicrously expensive (and currently unreleased as far as I know) 800mm super-tele on a full-frame body. I’d love to have a FF body for the wide-angle shots without needing specialty EF-S lenses and of course the additional image quality, but it’s the really long lenses that cost the big bucks and at this point in my career it’s not worth the extra money for the image quality gained (discounting the additional features only available on FF bodies, of course).

  20. Camera buffers will be in the gigabytes.

    The mechanical shutter will be removed.

    This will allow the camera when in powered on / standby mode to be constantly viewing the area and when the photographer presses the shutter it can be set to record 5 frames from the previous second so you never miss a photo.

    Liveview will be always instead of selected.

  21. Camera buffers will be in the gigabytes.

    The mechanical shutter will be removed.

    This will allow the camera when in powered on / standby mode to be constantly viewing the area and when the photographer presses the shutter it can be set to record 5 frames from the previous second so you never miss a photo.

    Liveview will be always instead of selected.

  22. The questions is…when will start to see some of this future ‘magic’ on the market. We are beginning to see more in the way of full frame options, as well as the high ISO capability, but not yet anything on the increased dynamic range front. Looking forward to that.

    Great post, Scott.

  23. The questions is…when will start to see some of this future ‘magic’ on the market. We are beginning to see more in the way of full frame options, as well as the high ISO capability, but not yet anything on the increased dynamic range front. Looking forward to that.

    Great post, Scott.

  24. I’m sort of curious if there is a reason the X3 technology from Foveon http://www.foveon.com/ has not had a bigger impact since it was developed? Essentially it utilizes the principle that red, green and blue penetrate silicon to different depths and embeds the sensors at appropriate levels of silicon. This way each pixel can have a red green blue value. Their point is one of their sensors captures triple the data of a standard sensor. It also seems like you would be able to get the function of three sensors on one.
    Is there some reason this is not a good technology? Maybe it does not work well in low light??
    I know they seem to have had problems getting this into any major camera’s, Sigma makes a couple but they don’t seem to be to popular. The technology seems like a pretty good approach at outside the box thinking.

    When I first read about this a few years ago I thought, well there it is the next big thing in digital cameras but it just hasn’t happened. Might that change?

  25. I’m sort of curious if there is a reason the X3 technology from Foveon http://www.foveon.com/ has not had a bigger impact since it was developed? Essentially it utilizes the principle that red, green and blue penetrate silicon to different depths and embeds the sensors at appropriate levels of silicon. This way each pixel can have a red green blue value. Their point is one of their sensors captures triple the data of a standard sensor. It also seems like you would be able to get the function of three sensors on one.
    Is there some reason this is not a good technology? Maybe it does not work well in low light??
    I know they seem to have had problems getting this into any major camera’s, Sigma makes a couple but they don’t seem to be to popular. The technology seems like a pretty good approach at outside the box thinking.

    When I first read about this a few years ago I thought, well there it is the next big thing in digital cameras but it just hasn’t happened. Might that change?

  26. I agree with Kent, the way we see digital censors now is very similar to film, but that will change as digital floats away from its conventional design. Soon the shutter will be stripped from dslrs and with that, individual pixel exposures will come into play meaning every part of an image will be perfectly exposed with just one shot; instead of having the whole censor exposed at one iso and one speed.

  27. “g. Megapixel madness will have finally stopped at about 100 megapixels with most cameras offering sensors in the 25-50 megapixels range.”

    No doubt the madness will only have waned after the release of the “Canon 1Ds MkInfinity” 1 Terapixel DSLR… Offering a blistering 3 fpd (frames per day) shutter speed, liquid nitrogen cooled 512 bit Digic Dios processor….and a drive-in movie screen to view your treasured memories…but only at night… :cool:

    Actually, what I’m really desperate to see is LCD monitors make similar leaps in pixel density as sensors. I want something crazy like an 8640 x 5400 resolution monitor that fits in the same real estate as a 24″ widescreen – that way I won’t neeed an LCD the size of the side of a barn view images at native 1:1 resolution.
    Sure, an LCD with that density would be pretty expensive since it wouldn’t be a mass consumer product…well, not to start anyway. But once people saw the end of things like pixellated text and awful “smooth fonts” in the web, Word, e-mail and OS’s, they’d be on board and the price would fall.
    Maybe all us photogs should start petitioning the three LCD panel manufacturers (and video card mfg) to let ‘em know there’s a demand for higher res/higher TFT density monitors…displays that get all the image on screen without interpolating it, and aren’t so big that you get whiplash after a few hours image editing. :D

  28. “g. Megapixel madness will have finally stopped at about 100 megapixels with most cameras offering sensors in the 25-50 megapixels range.”

    No doubt the madness will only have waned after the release of the “Canon 1Ds MkInfinity” 1 Terapixel DSLR… Offering a blistering 3 fpd (frames per day) shutter speed, liquid nitrogen cooled 512 bit Digic Dios processor….and a drive-in movie screen to view your treasured memories…but only at night… :cool:

    Actually, what I’m really desperate to see is LCD monitors make similar leaps in pixel density as sensors. I want something crazy like an 8640 x 5400 resolution monitor that fits in the same real estate as a 24″ widescreen – that way I won’t neeed an LCD the size of the side of a barn view images at native 1:1 resolution.
    Sure, an LCD with that density would be pretty expensive since it wouldn’t be a mass consumer product…well, not to start anyway. But once people saw the end of things like pixellated text and awful “smooth fonts” in the web, Word, e-mail and OS’s, they’d be on board and the price would fall.
    Maybe all us photogs should start petitioning the three LCD panel manufacturers (and video card mfg) to let ‘em know there’s a demand for higher res/higher TFT density monitors…displays that get all the image on screen without interpolating it, and aren’t so big that you get whiplash after a few hours image editing. :D

  29. We’ll pop in a bank of several 1TB miniSD cards.

    Option to write photos in real time to remote hard drive via built in wireless modem.

    Will the whole mirror system become obsolete? Seems like it could.

    DSLRs that capture super high quality video? For sure. (Don’t know if they will be able to call them SLR’s though)

    The industry will always want to keep a separation between “consumer” and “pro” level equipment. I’m sure they will also be very careful about anything that disrupts or dilutes devotion to existing product line concepts. How will they transition in new technologies? I think it will be interesting to see whether the innovations that are coming trickle down from the “pro” category, or trickle up from the “consumer” category. Maybe a little of both.

    One thing seems certain to me, the first point and shoot that has 0 shutter lag is going to have sales like no other. Once the rest follow suite we’ll see a serious decline in DSLR sales. Would it take a non Canon/Nikon to kill that golden goose? Seems like it could.

  30. We’ll pop in a bank of several 1TB miniSD cards.

    Option to write photos in real time to remote hard drive via built in wireless modem.

    Will the whole mirror system become obsolete? Seems like it could.

    DSLRs that capture super high quality video? For sure. (Don’t know if they will be able to call them SLR’s though)

    The industry will always want to keep a separation between “consumer” and “pro” level equipment. I’m sure they will also be very careful about anything that disrupts or dilutes devotion to existing product line concepts. How will they transition in new technologies? I think it will be interesting to see whether the innovations that are coming trickle down from the “pro” category, or trickle up from the “consumer” category. Maybe a little of both.

    One thing seems certain to me, the first point and shoot that has 0 shutter lag is going to have sales like no other. Once the rest follow suite we’ll see a serious decline in DSLR sales. Would it take a non Canon/Nikon to kill that golden goose? Seems like it could.

  31. Interesting, but I doubt all will be full frame. If they can build an smaller sensor with a high pixel count and great low noise performance why make it bigger? It is easier to make a great lens to cover a smaller sensor area than a one for a large sensor. And so that lens will be cheaper and lighter.

    There is nothing magical about the size of a full frame sensor. It doesn’t match proportion to most commonly used print sizes. It is simply a legacy technology. The first 35mm still cameras were designed as such because of the available of movie film stock available at the time. If that stock had been wider or narrower, we would all be shooting a different “full frame” sensor now.

    Advancing technology can provide a whole new way of shooting. Perhaps what is really needed is a new camera designed from the ground (super high performing sensor) up. I love my dSLR and I also love my Canon G9. I would really love a camera even smaller than the G9 with better low light performance and more pixels than any dSLR available today.

  32. Try to keep on point here folks. We’re talking about sensors. Not glass, shutters, etc.

  33. Try to keep on point here folks. We’re talking about sensors. Not glass, shutters, etc.

  34. Oops, didn’t mean to hijack the thread.

    Good point though Jim Buttitta. Why not make sensors even bigger then “full size”? EF could be the new crop increment and a whole new line of “plus sized sensor” lenses could be made.

  35. Oops, didn’t mean to hijack the thread.

    Good point though Jim Buttitta. Why not make sensors even bigger then “full size”? EF could be the new crop increment and a whole new line of “plus sized sensor” lenses could be made.

  36. Eric Stern said: “I think there will continue to be a market for crop sensor bodies though – the “free” extra focal length you get as a result is great when you need a long lens.”

    There will be no “free” extra focal length if technology allows for 35mm frame sensors with the same pixel density as APS-C ones. Which is what would result once technology reaches the point that Scott is suggesting.

    In that instance a shot taken on a smaller sensor camera would be identical to one taken with the same lens on a body containing a 35mm frame sensor which had then been cropped to the same resolution.

    You can even make that case with current technology, that a cropped shot taken with a 1DsIII will give better results than using the same lens on a 40D where the subject fills the frame.

    Although I agree with George Sinos, that there will still be a place for APS-C sensors.

    Some people want smaller and lighter cameras and lenses. And more cynically, maybe, it will be more profitable for manufacturers to have such products. You start with a point and shoot and (for want of a misappropriate term) when you outgrow it you get a ‘smaller format’ DSLR. Then when you want something more you can get a 35mm format DSLR.

    Besides which, if technology does progress in the way laid out then it would make medium format sensors just as affordable to the masses. So why not have one of those for your high quality work, and a handy lightweight ‘smaller format’ camera for general walkaround use?

    Michael.

  37. Eric Stern said: “I think there will continue to be a market for crop sensor bodies though – the “free” extra focal length you get as a result is great when you need a long lens.”

    There will be no “free” extra focal length if technology allows for 35mm frame sensors with the same pixel density as APS-C ones. Which is what would result once technology reaches the point that Scott is suggesting.

    In that instance a shot taken on a smaller sensor camera would be identical to one taken with the same lens on a body containing a 35mm frame sensor which had then been cropped to the same resolution.

    You can even make that case with current technology, that a cropped shot taken with a 1DsIII will give better results than using the same lens on a 40D where the subject fills the frame.

    Although I agree with George Sinos, that there will still be a place for APS-C sensors.

    Some people want smaller and lighter cameras and lenses. And more cynically, maybe, it will be more profitable for manufacturers to have such products. You start with a point and shoot and (for want of a misappropriate term) when you outgrow it you get a ‘smaller format’ DSLR. Then when you want something more you can get a 35mm format DSLR.

    Besides which, if technology does progress in the way laid out then it would make medium format sensors just as affordable to the masses. So why not have one of those for your high quality work, and a handy lightweight ‘smaller format’ camera for general walkaround use?

    Michael.

  38. “The sensors of the future will al be full-frame. The technology will exist to make these things for pennies, removing the need for cropped sensors.”

    What is magic about the particular dimensions of “full-frame”? Why not smaller for more compact cameras? Why not larger for easier heat dissipation?

    Honestly, I don’t see why we as an industry would migrate back to full frame, or if we were to move in that direction what would stop us at the 35mm frame size rather than continuing on to a larger frame (say, 1.5x “35mm” on a diagonal).

  39. I don’t see going to a single sensor size any more than I see going to a single mount. I think it may go the other way. Smaller sensors for smaller systems with an advantage of having longer range with smaller lenses. The 35mm/full frame sized sensors which do everything well. Slightly larger, but more versatile. And I could see a new, larger sized sensor that would put the squeeze on medium format camera systems. They would have a wide angle advantage over the other systems and would likely cost less than medium format systems.

    Heck, maybe someone could even come up with a round sensor. It would take full advantage of the whatever lens size it is designed for and will allow you more control of your own picture dimensions. After all, there are many different print sizes and you are going to be doing some cropping if you are printing no matter what shape the sensor is.

  40. I don’t see going to a single sensor size any more than I see going to a single mount. I think it may go the other way. Smaller sensors for smaller systems with an advantage of having longer range with smaller lenses. The 35mm/full frame sized sensors which do everything well. Slightly larger, but more versatile. And I could see a new, larger sized sensor that would put the squeeze on medium format camera systems. They would have a wide angle advantage over the other systems and would likely cost less than medium format systems.

    Heck, maybe someone could even come up with a round sensor. It would take full advantage of the whatever lens size it is designed for and will allow you more control of your own picture dimensions. After all, there are many different print sizes and you are going to be doing some cropping if you are printing no matter what shape the sensor is.

  41. There are only three things I am sure of on the list – the rest is conjecture based on my 35 years dealing with the camera companies. The full frame sensor is one of the three I am certain about. Then again, I have access to information most of you do not. In five years you can refer back to this article :)

  42. There are only three things I am sure of on the list – the rest is conjecture based on my 35 years dealing with the camera companies. The full frame sensor is one of the three I am certain about. Then again, I have access to information most of you do not. In five years you can refer back to this article :)

  43. @Tom Dibble, The magic of the present ‘full frame’ 35mm sensor size has to do with the fact that millions of lenses over the past several decades have been sold with am image circle that covers a 24x36mm frame. Shrinking the frame to an APS-C size is no problem as the existing lenses will work. Increasing the frame size above 24×36 will require a whole new set of lenses with a larger image circle to cover the frame. These lenses will be larger, heavier, and more expensive.

    There IS already several larger sensor formats that are based on the old film medium formats. 645 and 2 1/4 square are both standards and there ARE digital backs with sensors that fit these formats available today from manufacturers such as Leaf, Phase One, Mamiya that fit on existing firm medium format cameras and use the existing lenses made for this format. At the present state of the art, these larger formats are expensive. I believe the lowest cost MF digital back available today, (not including the camera or lens) is around $12-$15K.

  44. @Tom Dibble, The magic of the present ‘full frame’ 35mm sensor size has to do with the fact that millions of lenses over the past several decades have been sold with am image circle that covers a 24x36mm frame. Shrinking the frame to an APS-C size is no problem as the existing lenses will work. Increasing the frame size above 24×36 will require a whole new set of lenses with a larger image circle to cover the frame. These lenses will be larger, heavier, and more expensive.

    There IS already several larger sensor formats that are based on the old film medium formats. 645 and 2 1/4 square are both standards and there ARE digital backs with sensors that fit these formats available today from manufacturers such as Leaf, Phase One, Mamiya that fit on existing firm medium format cameras and use the existing lenses made for this format. At the present state of the art, these larger formats are expensive. I believe the lowest cost MF digital back available today, (not including the camera or lens) is around $12-$15K.

  45. Regarding Scotts comment about 100MP image sensors on a 24x26mm frame. I doubt that the number of pixels on a 35mm frame will go much past 36MP. The reasons are geometry and physics. As the pixel size decreases, the amount of photons that each photo site can capture also decreases for a given exposure. The signal to noise ratio drops as does dynamic range and you end up with noisy images with reduced dynamic range. The signal to noise ratio is approximately equal to the number of photons captured divided by the square root of the number of photos captured. So the S/N Ratio decreases linearly with the decrease in pixel size. Certainly camera electronics will be improved, but there is always noise inherent in the capture of photons, conversion to electrons, the moving of electrons and the A/D conversion.
    If ones goal is increased dynamic range and increased high ISO performance, shrinking the pixel size will hinder both of those goals.

    Also, the ability to design lenses that can actually resolve detail onto an area as small the pixels would be on such a sensor would be a difficult feat as well as costly. I doubt that the best lenses made today by Canon & Nikon can resolve much past 30-40MP based on what I have read on the topic.

    Here is a table of pixel pitch vs No of Mega Pixels on a 24x36mm sensor vs photon well depth
    8um 13.5MP 80,000 (Very close to the Canon 5D)
    7um 17.6MP 70,000 (Close to the Canon 1Ds Mk2)
    6um 24MP 50,000 (Very close to the Canon 1Ds Mk3)
    5um 34MP 35,000 (Near or at the resolving power of top end lenses)
    4um 54MP 20,000 (At or beyond the resolving power of top end lenses)

    Here are a couple of links to info on the topic:

    http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/

    http://www.outbackphoto.com/dp_essentials/dp_essentials_02/essay.html

  46. I might add that all of what I wrote above is based on current CCD & CMOS sensor technology. Some new sensor technology may come along that would allow all of the items on Scotts list to become reality. Probably not in 5 years though as that new technology has not yet been or has barely been invented and new technologies take time, probably a decade or so to develop into a commercial reality. I am certainly optimistic about what the future holds. In 10 to 20 years, we will look back and laugh at the present state of the art of digital photography.

    It is amazing to look back 10 years to 1998, a 1-2MP DSLR was the state of the art, with a horriblly large form factor and cost $30K A 1MP Point & shoot cost ~$1000. Image quality was no where what it is today.

  47. I might add that all of what I wrote above is based on current CCD & CMOS sensor technology. Some new sensor technology may come along that would allow all of the items on Scotts list to become reality. Probably not in 5 years though as that new technology has not yet been or has barely been invented and new technologies take time, probably a decade or so to develop into a commercial reality. I am certainly optimistic about what the future holds. In 10 to 20 years, we will look back and laugh at the present state of the art of digital photography.

    It is amazing to look back 10 years to 1998, a 1-2MP DSLR was the state of the art, with a horriblly large form factor and cost $30K A 1MP Point & shoot cost ~$1000. Image quality was no where what it is today.

  48. Duppose we have chips that can perform reasonably well at 128,000 ISO. They would be noisey, no doubt. The camera could limit ISO availability to 32,000 or so. Then when we call for a 1 second exposure at 32,000 the camera could take three 1/3 second exposures at a higher ISO and blend them internally.

    Granted there are situations where this would fail. It could work fine in others. This can be done now manually to reduce noise why not have the sensor be able to do it to artificially increase the ISO?

  49. Duppose we have chips that can perform reasonably well at 128,000 ISO. They would be noisey, no doubt. The camera could limit ISO availability to 32,000 or so. Then when we call for a 1 second exposure at 32,000 the camera could take three 1/3 second exposures at a higher ISO and blend them internally.

    Granted there are situations where this would fail. It could work fine in others. This can be done now manually to reduce noise why not have the sensor be able to do it to artificially increase the ISO?

  50. There is a similar limitation in getting more effective bits too. A camera can claim 14 bits exposure data, but the effective is certainly less.

  51. There is a similar limitation in getting more effective bits too. A camera can claim 14 bits exposure data, but the effective is certainly less.

  52. How about sensors that are convex in shape, similar to the inside of the human retina? The thinking I have is that this shape makes the distance from the lens to the sensor exactly the same regardless of the area on the frame (ie. top left, center and bottom right are all equidistant to the rear element.

    Essentially matching the contour of the rear element on the lens with the contour of the sensor.

    Today you need the lens to bend the light so the picture does not stretch at the boundaries of the image. This effect can be seen on fish-eye lenses.

    A nice side-effect is that you can pack more sensor into less space, due to the hollow shape.

  53. How about sensors that are convex in shape, similar to the inside of the human retina? The thinking I have is that this shape makes the distance from the lens to the sensor exactly the same regardless of the area on the frame (ie. top left, center and bottom right are all equidistant to the rear element.

    Essentially matching the contour of the rear element on the lens with the contour of the sensor.

    Today you need the lens to bend the light so the picture does not stretch at the boundaries of the image. This effect can be seen on fish-eye lenses.

    A nice side-effect is that you can pack more sensor into less space, due to the hollow shape.

  54. As Jeff pointed out, as pixel density increases, so does noise. I firmly believe that Scott is right about the “full frame” thing, if for no other reason than for marketing. As pixel density increases, the advantage of larger pixels will vanish.

    I’m not sure I agree that an APS/C-sized crop from a full frame sensor will give a higher quality image than an APS/C sensor if all other things are equal. If the sensors are of equivalent design, that is, if all other things *are* equal, there should be no difference between the images.

    I’m not sure how that three-ccd thing is going to work. Video cameras use splitters to get three images – talk about a big, bulky camera, with 3 full-frame sensors. The Foveon X3 produces gorgeous color, but it’s limited to 4mp (2640×1760 – they call it 14mp because there are three – one for each color – of each one of those pixels) in the current incarnation, and from my visits to pixel-peeper.com, seems to have some color fringing and sharpness issues compared to Canon, Nikon, Pentax, and friends. The Bayer sensor is a pretty doggone good compromise, as nearly as I can tell.

  55. The Foveon sensor has very deep pixel wells due to the fact that there are three layers of photo sensitive silicon all stacked on top of one another. This makes it difficult for each photo site to collect photos that fall on it at an angle, especially the lower layers. One solution to this shortcoming is to put a micro lens over each pixel. To optimize this solution the position with respect to the center of the pixel and the radius of each lens would have to vary as a function of the distance from the center of the sensor. This is in fact the solution that Leica used on the sensor on their M8 to allow a large sensor to be placed very close to the rear lens element. Unfortunately this is expensive and tricky to make. The addition of micro lenses also limits how small the pixel pitch can be. Both of these factors, deep pixel wells and micro lens restrictions are most lilely why Sigma has not produced a higher resolution chip.

    @Steve, Making the sensor spherical is an interesting idea, but I have no idea how a non flat silicon device could be made. Selecting the radius would also be a compromise as each wavelength of light throughout the visible spectrum would optimally require a different radius. Lens optics solved this problem long ago. Having many movable optical components within a lens, it can come close to directing each wavelength of light to the appropriate location on the flat image sensor.

  56. The Foveon sensor has very deep pixel wells due to the fact that there are three layers of photo sensitive silicon all stacked on top of one another. This makes it difficult for each photo site to collect photos that fall on it at an angle, especially the lower layers. One solution to this shortcoming is to put a micro lens over each pixel. To optimize this solution the position with respect to the center of the pixel and the radius of each lens would have to vary as a function of the distance from the center of the sensor. This is in fact the solution that Leica used on the sensor on their M8 to allow a large sensor to be placed very close to the rear lens element. Unfortunately this is expensive and tricky to make. The addition of micro lenses also limits how small the pixel pitch can be. Both of these factors, deep pixel wells and micro lens restrictions are most lilely why Sigma has not produced a higher resolution chip.

    @Steve, Making the sensor spherical is an interesting idea, but I have no idea how a non flat silicon device could be made. Selecting the radius would also be a compromise as each wavelength of light throughout the visible spectrum would optimally require a different radius. Lens optics solved this problem long ago. Having many movable optical components within a lens, it can come close to directing each wavelength of light to the appropriate location on the flat image sensor.

  57. Your list doesn’t surprise me at all if I think about it. It’s more a question of which manufacturer will start with those improvements.
    Besides that, you might even have forgotten some things. And as with all things in life; We’ll see, or we won’t.

  58. Your list doesn’t surprise me at all if I think about it. It’s more a question of which manufacturer will start with those improvements.
    Besides that, you might even have forgotten some things. And as with all things in life; We’ll see, or we won’t.

  59. a. Making a small chip will always be cheaper than a big one. So full frame I think will always be relegated to the high end. Big sensor means bigger lens, compact cameras will always have small sensors.

    b. Price always goes down, non news.

    c. 11+ stops in 5 years. No. Maybe 10-15. Maybe.

    d. Yah ISO has been going up a stop every generation or two. That will make sense.

    e. This one will come around about the time of 11+ stops of dynamic range. But I think floating point will be more probable.

    f. For most companies that don’t have stabilized lenses already that’s probably true. But why would Canon and Nikon change their way? Not in the next 5 years.

    g. I think megapixel madness is slowing as we speak. And even if it grows twice as fast we won’t 100Mp in 5 years. Remember we are only talking 3 or so generations over 5 years.

    h. Not in SLRs! The bodies would have to get huge! (the size of high end camcorders) to fit 3 full frame sensors and all the prisms needed.

    i. I can see this one. Maybe not a coating but some kind of electrostatic screen so the sensor actually repels dust.

    j. Maybe in some applications, but I don’t think it will be common. Especially in DSLRs.

  60. a. Making a small chip will always be cheaper than a big one. So full frame I think will always be relegated to the high end. Big sensor means bigger lens, compact cameras will always have small sensors.

    b. Price always goes down, non news.

    c. 11+ stops in 5 years. No. Maybe 10-15. Maybe.

    d. Yah ISO has been going up a stop every generation or two. That will make sense.

    e. This one will come around about the time of 11+ stops of dynamic range. But I think floating point will be more probable.

    f. For most companies that don’t have stabilized lenses already that’s probably true. But why would Canon and Nikon change their way? Not in the next 5 years.

    g. I think megapixel madness is slowing as we speak. And even if it grows twice as fast we won’t 100Mp in 5 years. Remember we are only talking 3 or so generations over 5 years.

    h. Not in SLRs! The bodies would have to get huge! (the size of high end camcorders) to fit 3 full frame sensors and all the prisms needed.

    i. I can see this one. Maybe not a coating but some kind of electrostatic screen so the sensor actually repels dust.

    j. Maybe in some applications, but I don’t think it will be common. Especially in DSLRs.

  61. [...] Quest » Blog Archive » The End Of A Black-And-White World: Color PhotographyTWIP » The Future of Digital Sensors – TWIP Tags digital photography photography tip digital photo photography light wedding photography [...]

  62. [...] Quest » Blog Archive » The End Of A Black-And-White World: Color PhotographyTWIP » The Future of Digital Sensors – TWIP Tags digital photography photography tip digital photo photography light wedding photography [...]

  63. As a real estate photographer I’m forced to rely on HDR instead of expensive and slow to set up strobes for shooting complicated interior photos. I’m hoping hoping for expanded HDR options such as:
    1. In camera HDR processing
    2. A sensor that you could have a bracketed ISO feature. This would mean that for a single shot, so you would have 3-6 exposures each at a different ISO so the exposure would be bracketed without having to take multiple pictures.

Comments are closed.

About scottbourne

Founder of Photofocus.com. Retired traveling and unhooking from the Internet.

Category

Opinion, Technique & Tutorials

Tags