• This topic has 12 replies, 9 voices, and was last updated 12 years ago by Bez.
Viewing 13 posts - 1 through 13 (of 13 total)
  • Geeky camera question
  • molgrips
    Free Member

    So, we know that high megapixels on a compact camera is a bad thing because it results in noisy images that require loads of NR. The question is, if I reduce the MP count of the final image, does this help?

    I suppose if it’s averaging the values of surrounding pixels it would improve matters – ideally though you’d want to reduce to 25% of the highest number so that four pixels get averaged out.. maybe?

    Also if you’ve got no access to the NR settings then I suppose it’ll still apply a high NR setting anyway, which could still mess up the image.

    5thElefant
    Free Member

    Yes it does exactly what you suggest. Or maybe it doesn’t but that’s the way I understand it and the end result is you can’t see the noise anymore (as much).

    molgrips
    Free Member

    Right, I have yet to upload any pics from my mju so will try it out 🙂 I think there’s no point at all in leaving it on the 14MP setting.

    Fresh Goods Friday 696: The Middling Edition

    Fresh Goods Friday 696: The Middlin...
    Latest Singletrack Videos
    GrahamS
    Full Member

    It won’t help – you’ll just get a soft looking image with slightly blurrier noise.

    You’d be better off leaving it as full res and running it through a proper noise reduction filter.

    Three_Fish
    Free Member

    The question is, if I reduce the MP count of the final image, does this help?

    No. The noise will still be there from the original image.

    UrbanHiker
    Free Member

    If your going to reduce the resolution in post production, would it be sensible to take it with that resolution in the 1st place? Does the camera then average the 4 pixels on the ccd?

    Not speaking with any authority, just thinking out loud.

    GrahamS
    Full Member

    Does the camera then average the 4 pixels on the ccd?

    Doesn’t matter. The fundamental problem is with the size of individual photosites on the sensor. When you try to cram in millions of them on an already small compact sensor then you have to make them incredibly tiny – so they are subject to noise.

    Think millions of tiny little buckets try to catch water. Any little splash or overflow ends up filling adjacent buckets. If the buckets were bigger (larger photosites) then the same thing happens but contributes less to the amount of water (light) in the bucket so is less of a problem.

    Taking the average of 1 noisy and 3 good pixels will still look noisy, but a little blurry.

    grum
    Free Member

    So, we know that high megapixels on a compact camera is a bad thing because it results in noisy images that require loads of NR.

    Is it actually that the images are noisy though – I mean when viewed/printed at normal sizes? Or is it just that they look noisy when you zoom into them/blow them up large because the camera is cramming in more MP than the sensor can make good use of? If you see what I mean.

    molgrips
    Free Member

    You’d be better off leaving it as full res and running it through a proper noise reduction filter.

    No RAW feature on this camera and I can’t find any way of taking NR off.

    Doesn’t matter. The fundamental problem is with the size of individual photosites on the sensor.

    I disagree – noise is random variation around what the true value of the pixel should be. More or less. So averaging four into one would get closer to the true value of incident light on that part of the sensor I reckon.

    If your going to reduce the resolution in post production, would it be sensible to take it with that resolution in the 1st place? Does the camera then average the 4 pixels on the ccd?

    That’s what I’m talking about – doing it in camera. I dunno how it shrinks the images or at what stage. There doesn’t seem to be a 3.5mp setting which would be 1/4 of the original image.

    oliverd1981
    Free Member

    I guess a good post processing noise filter discards wayward pixels where as the camera won’t have the ooomph to do sophisitcated analysis on the fly. At lower resolutions it will sample less photosites (thats why you can take more pictures in burst mode at lower resolutions, it’s not processing as much data)

    Don’t some high iso settings reduce the resolution and use the averaging to their advantage?

    Bez
    Full Member

    Averaging is also known as pixel binning. It is a valid NR technique if you’re happy to lose resolution. Some cameras do it, others don’t, try your luck and compare results – take the same photo in various camera settings and then try downscaling them to matching sizes.

    If you really want to reduce noise, try stacking (averaging multiple exposures) – though of course that only works if you have a static scene and a static camera.

    Conqueror
    Free Member

    Basically what E said (GrahamS)

    Noise also increases as pixel size decreases

    http://www.dpreview.com/learn/?/key=noise

    Bez
    Full Member

    The words ‘pixel’ and ‘photosite’ often get a little confused. The former is a part of an image, the latter a part of a sensor.

    Noise also increases as pixel size decreases

    Noise in the captured data increases as photosite size decreases. We’re talking about images coming from one sensor here. The fact that it’s a small sensor is totally immaterial. When we’re comparing full-resolution images with reduced-resolution images, the photosite size is the same.

    If you average down the pixels to a lower resolution then you’re averaging out the random noise, and the noise (in the resulting image) will thus reduce.

    This is actually in agreement with the statement you quote, since if you downsize to 1/4 of the original resolution (ie halve both width and height of the image) a pixel in the final image now represents four times the physical area – ie you have larger pixels, thus less noise.

Viewing 13 posts - 1 through 13 (of 13 total)

The topic ‘Geeky camera question’ is closed to new replies.