The Joy of Pixel Density

Started Jul 13, 2008 | Discussions thread
Tristan Cope Senior Member • Posts: 1,141
Density limits

I've been following the discussions about pixel size in recent threads with some interest. If I understand correctly you saying that although (for a given level of technology) for a fixed number of pixels larger sensors are better than smaller sensors, for a given sensor size more pixels is always better because (a) you get better resolution and (b) you can reduce noise by binning pixels.

(to clarify by pixel we are referring to sensor element rather than image element)

The counter argument goes along the lines of: The smaller pixels have lower DR because of their smaller well size and they have more noise.

To which your response is that the smaller well size of the smaller pixels doesn't matter for DR because the same number of photons are spread between more pixels. If the larger pixel doesn't saturation then the collection of smaller ones won't either. As for noise, you can downsample the higher resolution image and average out the noise.

All of this seems to make sense to me, and you've backed it up with some samples. Again, if I understand correctly the samples compare an image from a small sensor compact with a crop from a larger sensor camera such that the two images displayed come from the same physical area of sensor but with different numbers of pixels (thus providing a surrogate to comparing two sensors of the same size with differening numbers of pixels).

I know that images from my DSLRs are signifcantly better than those from my compact cameras with a similar number of megapixles. The difference at low ISO is only visible when viewing at large sizes (eg viewing at 100% on my monitor or printing at larger sizes). At higher ISO the difference becomes much more obvious. But of course the DSLR sensor is many times larger than the compact sensor.

For sensors of equal size, many claim that squeezing extra MP onto a small sensor makes the image quality worse (in terms of noise and DR) but I expect they are viewing the images at 100% on a computer monitor rather than viewing (or printing the images at the same size). I suspect though that there may be individual examples from some manufacturers where an increase in the number of pixels from one model to the next has not been beneficial.

However, presumably there must be a limit to how small the pixels can go? The sensor elements need to receive photons in reasonable numbers in order for us to be sure there is a signal there. When photon numbers get very low (say down to one or two per pixel for the exposure) their behaviour is unpredictable. At around one photon per pixel it is impossible to distinguish between signal and noise. At this point binning pixels won't help because the signal cannot be extracted from the noise.

As pixel size gets smaller, shot noise becomes more of a problem - especially for low light conditions. Don't you think small sensor compacts (especially 1/2.8") with 10+ MP are already reaching that point?
I just like the pictures.

 Tristan Cope's gear list:Tristan Cope's gear list
Sony RX100 Sigma SD14 Panasonic Lumix DMC-GX1 Fujifilm X-Pro1 Canon EOS 5D Mark III +1 more
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow