My Nikon D5600 sensor can deliver 24MP but I configured it to just use 6MP. At which step in the process does the camera "scales" the images down?

Maybe already at sensor level? By calculating an 'average' of the signal of four neighboring pixels? Or by using only every fourth pixel?

Before or after noise reduction? Before or after sharpening?

Or do they just scale the image down, just before writing it to the SD card?

What does it do to the image quality? Will I see less noise? Will higher ISO settings be possible because of less noise?

PS: The Bayer filter was mentioned in the comments. When I think about it, that makes it even more complicated. Or interesting:

This means, to some degree, they have to scale up the image/color signal to reach that 24MP?

Then I am even more curious if, when shooting with 6MP, they first scale up (by interpolating) the data to 24MP and later scale it down again or if they use another algorithm for 6MP (i doubt) without interpolating? And if such an alternative 6MP algorithm would produce crispier results?

  • 2
    \$\begingroup\$ Welcome along, good question, and one that interests me... \$\endgroup\$ Commented Aug 23, 2017 at 8:53
  • 1
    \$\begingroup\$ Another option, sometimes used for video, is to only use the central portion of the sensor (which provides a higher crop factor, of course). \$\endgroup\$
    – ths
    Commented Aug 23, 2017 at 12:15
  • \$\begingroup\$ Related: Does it give higher quality to shoot in low-res mode in-camera, or to downsize high-res photos later? \$\endgroup\$
    – mattdm
    Commented Aug 23, 2017 at 13:43
  • \$\begingroup\$ I believe it only scales at jpeg generation, since it does not offer low res raw. And i'd hope it does it before sharpening, but i have no definitive answer. \$\endgroup\$
    – ths
    Commented Aug 24, 2017 at 19:31
  • 1
    \$\begingroup\$ It's possible that every camera maker offering this feature does it differently. \$\endgroup\$ Commented Aug 24, 2017 at 21:31

1 Answer 1


Commonly cameras capture data using the full sensor which gives full-resolution RAW data which is converted into an image and then downscaled. As a matter, if you select to shoot JPEG+RAW, you do still get a full-resolution RAW file and a scaled down JPEG.

Your proposed using every 4th pixel or average of neighboring pixels is unlikely since it does not work well with Bayer filters. At best, you would get an image with heavy aliasing and strong moire. This is a technique which has been criticized when used for video where lower-bandwidth is needed.

Camera makers have tried different things. The case of quarter resolution is easy since the Bayer filter repeats a 2x2 grid of Red, Green, Green, Blue subpixels. To get exactly quarter resolution, a full-color image can be obtained by just using the Red and Blue directly, plus averaging the two Green. You still need to get the same amount of output from the sensor.

Fuji notably implemented a sensor which allowed them to get the quarter resolution image right off the sensor by using a pattern of 45°-rotated colors which they called EXR Color Array. This makes each two subpixels of the same color adjacent and can be read at the same time which reduced noise and bandwidth. The official Fuji page seems gone but here is an article about it.

Noise reduction could be applied either before or after but since reducing resolution also reduces how much noise is apparent, it will be probably done later. Sharpening needs to be done as a last step due to the fact that downscaling makes images softer.

Some cameras also offer quarter resolution RAW. I suspect they do it with a special downscaling to the RAW data as other ways would require additional circuitry on-sensor which would have impact to noise, heat and complexity.


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.