0
\$\begingroup\$

AFAIK, image sensors use demosaicing algorithms to reconstruct virtual image pixel from different sub-pixels. Does it mean, that on some cameras it could be profitable to use lower-resolution modes to get better out-of-the-box picture?

I shoot show-jumping on Canon 80D indoors and not so good in using raw-therapee to get really good results from RAWs (and sometimes there are just bugs, ex. I see no difference when applying some noise reduction). And on JPEGs image is quite blurry on HighISO even on my FullHD monitor due to noise, so there is absolutely no sense in additional megapixels. With new 90D on the way and increased megapixel size it becomes even more actual.

So whether I will see better picture on my monitor using in-camera lower resolution compared to L-size JPEG that is converted to FullHD automatically by image viewer or by converters of social networks?

\$\endgroup\$
5
  • \$\begingroup\$ What makes you think demosaicing and noise might be related? \$\endgroup\$
    – xiota
    Sep 23, 2019 at 14:13
  • \$\begingroup\$ Can you add a sample image that you think is "quite blurry... due to noise"? \$\endgroup\$
    – xiota
    Sep 23, 2019 at 14:16
  • \$\begingroup\$ What do you mean by "it becomes even more actual"? \$\endgroup\$
    – xiota
    Sep 23, 2019 at 14:17
  • \$\begingroup\$ Have you tried using S or M settings? What were the results? \$\endgroup\$
    – xiota
    Sep 23, 2019 at 14:19
  • \$\begingroup\$ It's worthy to note that they are not sub-pixels. Each pixel is either red, green, or blue and the algorithm simply looks at neighboring pixels to determine the real color of each pixel. Also, you'll get much better feedback by providing a sample image with EXIF info. If you're seeing no difference when applying noise reduction then you're probably doing something wrong. \$\endgroup\$
    – Mike Dixon
    Oct 23, 2019 at 12:25

2 Answers 2

1
\$\begingroup\$

It is likely that scaling down the image in post-production will remove as much noise as the in-camera processing. But in one case all you have is the low-res picture, and in the other you (or someone else with more experience with raw files or image editing) can go back to the full image if necessary.

Of course the 90D may have ore pixels, but:

  • for the same ISO you could have less noise
  • if you scale down to the same absolute size, you will reduce noise even more
\$\endgroup\$
0
\$\begingroup\$

AFAIK, image sensors use demosaicing algorithms to reconstruct virtual image pixel from different sub-pixels. Does it mean, that on some cameras it could be profitable to use lower-resolution modes to get better out-of-the-box picture?

Unfortunately not. I think what you are asking is can I get away without using the demosaicing algorithm if I use a lower-resolution mode and thus get a more accurate reading of the light at the sensor level and the answer is no. Virtually every digital camera has a physical layer over the sensor (with the strange exception of the monochrome-only Leica M Monochrom digital camera) to introduce colour information. This is done by blocking out different colours at different pixel sites and then the demosaicing algorithm averages out the colours to produce a colour image. This processing is done before any in-camera image sizing, so specifying a JPEG S or M size will not make a difference to the image quality. If you specify mRAW or sRAW the same thing happens in Canon digital cameras - they downsample the full resolution raw image data. There is a technique at the hardware level to use pixel binning as a way to reduce noise and resolution by averaging pixel data but that is not the method used by Canon.

For more details on mRAW and sRAW, see: https://photographylife.com/sraw-format-explained

\$\endgroup\$

Your Answer

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

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