Why no pocket digicams with CMOS sensor?

Discussion in 'Mirrorless Digital Cameras' started by stevierose, Jun 11, 2005.

  1. I wonder why Canon or someone else hasn't made a non SLR digicam with
    a CMOS sensor? I have loved my Canon DSLR's ever since the D30 because
    they produce nearly noiseless images even at high ISO settings. I have
    never liked any of the available smaller digicams because they all
    have terrible noise levels once you get above ISO 100 setting. I think
    if Canon put even one of their older CMOS sensors (from the first
    Digital Rebel, or even the D30) into a smaller digicam so that it took
    decent photos at ISO 400-800 they would just take the whole market! Do
    you folks think it is a marketing decision to drive buyers to the more
    expensive DSLRs? Is it a lack of production capacity for CMOS sensors?
    I don't get it.
  2. Sensors from DSLRs are about 10-15 times the size as sensors from digicams. Even the older DSLR sensors, where the cost of development has been amortized, are a lot more expensive to produce because of it.
  3. There's no advantage. The difference between good CMOS and CCD sensors these days isn't enough to make a difference. You use whatever's cheapest and available to you.
  4. I think the noiselessness of the images has a lot more to do with sensor size than sensor
    type. The smaller digicams are not set up optically to take such a large sensor as the ones in
    the DSLRs.
  5. Canon's noiselessness has more to do with their amplification and processing of the image than sensor type. In fact, CMOS in inherently noisier than CCD.

    If canon were to put a DRebel sensor in a digicam, it would either need a fixed lens, a much bigger zoom (not a compact anymore) or a much slower lens. Remember that even 4 times zoom 35mm compacts have horribly slow lenses - like f/4.5-12 - to make them fit in such a small package. Why do you think you need to load them with ISO800 film to get them to not use the flash all the time!?

    That said, if Canon - or anyone else - would make an APS-C or Four-Thirds sensor powered digicam with fixed 28/2.8 lens they'd have me as a customer!
  6. There are several digital compact cameras with low noise at high ISO. The best one is the Fuji FinePix F10. Its performance is extraordinary even at ISO 1600. Oh and BTW its sensor is a CCD. ;-)
  7. Spearhead

    Spearhead Moderator

    I can't understand why anyone other than a camera designer would care about the sensor type. Cameras do what they do as a system. Either a camera does what you want or it doesn't.
  8. Karim,

    I think you go a little over the top when you say there are "several"- I haven't seen one yet that was decent at high ISO's. The Fuji you mention may be good compared to other P&S digicams- heck not too many even give you an option over ISO 400, but that's it.
  9. The Fuji Z1 has the same techonology as the Fuji F10 but uses a smaller 1/2.5" sensor.
  10. jbq


    Some really cheap cameras are using CMOS sensors.

    The problem with higher sensitivities and small sensors is as pretty simple: there's inherently noise in the incoming light, and at normal sensitivities that noise translates into image noise. For a given sensitivity and pixel size there's a minimum amount of noise that a perfect sensor will record, there's no way around that. The smaller the sensor, the bigger that noise.

    As for the CMOS vs CCD design, if I remember correctly CMOS designs require more per-pixel logic than CCD ones, therefore lowering the size of the photosites. On large pixels the difference is negligible, but on small pixels it's not the case.
  11. ...pretty simple: there's inherently noise in the incoming light,...
    Warning, Will Robinson, warning. Bad Science alert!
  12. jbq


    Brad: I'm looking forward to your counter-explanation.

    Personally, I'm satisfied with the observations that for middle sensitivities noise grows by about 1.4x per stop for a given camera, why when comparing cameras will different pixel sizes noise seems to grow inversely proportionally with the pixel linear dimension (and not the surface), or why the noise levels that get measured on the middle sensitivities of any sensor reasonably match the square root of the number of photons expected to hit each pixel.
  13. Images are not noisy. The sensor's ability to accurately detect/accumulate the number of
    photons does follow a Poisson statistical distribution. This is called shot noise. There are
    other sensor noise sources that are thermally generated, as well as overall signal chain
    (sensor - amplification - ADC) noise sources as well.
  14. Personally, I am waiting for the digital equivalent of something like the Leica CM-Ricoh GR1V-Contax T3-Rollei AFM35-whatever. I want a compact digital camera with aperture priority that delivers good images at ISO 400, with a sharp and fast lens.

    I used to be able to load Velvia slide film in my EOS and compact film camera; I still have my film EOS, waiting for something to replace the film compact for street shooting.

    Unfortunately, camera manufacturers are more interested in having 30 "scene modes" and megazooms, and really tiny optical viewfinders that make you squint.
  15. Actually Jean-Baptiste observations are scientifically well founded and the shot noise that Brad mentioned is inherently in the light, not in the sensor. The shot noise of the light contained in the captured signal varies as the square root (1.414), and it cannot be removed regardless of the sensor design.

    As to the differences due to pixel size, the larger pixels have better signal-to-noise performance which improves the image quality, the noise in a smaller pixel isn't necessarily worse, but the signal will be lower since the photosensitive area is less.
  16. Leica announced they aren't going to develop a digital camera for their M-System. They say that M-lenses are not designed for didtal sensors due to the lifght rays coming from them wouldn't hit the sensor quite perpendicularly as requested. That could be the reason for that some M-lenses are reported to do some vignetting on the new Epson camera. No need to say that the same lenses perform excellently on Leica film cameras.
  17. My daughter's $99 'no brand' focus-free digicam uses a CMOS; so is the one on my Nokia 7610.
  18. Jeff Spirer said "I can't understand why anyone other than a camera designer would care
    about the sensor type. Cameras do what they do as a system. Either a camera does what
    you want or it doesn't."

    I agree with you Jeff. What I want is a compact digital camera that can produce fairly low
    noise images at ISO 400 and 800 equivalent settings similar to the images produced by
    the Canon DSLR cameras. I (apparently) mistakenly assumed that this was a property of
    the CMOS sensors they use. I don't care about the sensor type, I care about the image
    qualtiy the camera produces.
  19. The biggest barrier to this kind of camera is the angle at which the light hits the sensor. APS-C DSLR's work as well as they do because the light from the lenses is hitting the sensor at fairly straight angles. With image sensors, the detector in the photosite resides down in a small pit. If the light is coming into the pit at too great an angle a shadow is created over the detector by the pit walls. This is probably the main reason why Leica used to say there would be no digital M until the Epson RD-1 came out.

    The Epson RD-1 is the first step in the direction towards a APS-C sensored compact digicam. Here they figured out how to design a system that lets the rear element get close to the sensor without vignetting too badly. It still vignettes, but the amount is usually pretty managable. Between advances in microlens designs and in-camera image processing systems I suspect we'll see an APS sensored compact within the next five years. That is unless someone figures out how to make 1/1.8 or 2/3 sensors produce images as clean and as detailed as DSLR's at ISO 1600.

Share This Page