0
\$\begingroup\$

As I have figured, the main cause for the intensity values in an image to be non-zero even in the absence of light is the dark current which propagates through the photodiodes, providing some charge accumulation which in turn creates a non-zero intensity level for pixels in the resultant image even when there is no light through the lens.

So, to alleviate this, the black level offset is subtracted from the intensity values of an image generated when exposed to light, so as to enable the proper dynamic range.

But, by virtue of manufacture, each pixel location which is basically a photodiode or a photosensitive material will have a different amount of dark current, which would imply a different correction value for each spatial pixel location.

That would mean storing a huge amount of data to identify the black level correction for each pixel. How is this issue dealt with then?

\$\endgroup\$
4
  • \$\begingroup\$ Answer on wikipedia. technically not that much data since hopefully the range of values is fairly small. \$\endgroup\$
    – xenoid
    Commented Jul 29, 2019 at 11:46
  • 1
    \$\begingroup\$ "That would mean storing a huge amount of data to identify the black level correction for each pixel." – No more "huge" than the size of a single raw image file. \$\endgroup\$
    – xiota
    Commented Jul 29, 2019 at 16:36
  • \$\begingroup\$ "So, to alleviate this, the black level offset is subtracted from the intensity values of an image generated when exposed to light, so as to enable the proper dynamic range." – Not necessarily. What's the "proper dynamic range" of total darkness? Are you referring to a specific camera model or a hypothetical scenario? \$\endgroup\$
    – xiota
    Commented Jul 29, 2019 at 16:37
  • \$\begingroup\$ @xenoid There's a difference between the amount of data and how compressible it is. An infinite stream of zeros is infinite data, but it's also infinitely compressible. \$\endgroup\$
    – xiota
    Commented Jul 29, 2019 at 16:43

1 Answer 1

1
\$\begingroup\$

Stuff like dark current is highly depending on environmental conditions like the current temperature distribution across the sensor. For exposure times where the dark current of photodiodes is likely to make a difference, a camera often immediately makes another photograph of the same duration with closed shutter in order to get a pixel and environmental condition dependent current black frame compensation value. Cameras of mine that do this display "Processing" during that time but there really is no noise reduction algorithm that would take that kind of time on current camera hardware, so it's obviously just the creation of a black frame compensation value with closed-shutter exposure.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ This is typically an option that can be turned on/off. It's labeled "Long exp NR" in the shooting menu for Nikons. The dark frame data is then subtracted from the image data before writing the image data to the card. So the final image file would technically be smaller if anything. \$\endgroup\$ Commented Jul 29, 2019 at 13:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.