3
\$\begingroup\$

Can someone help me visually identify Pixel vignetting (such as the one described here: https://photographylife.com/what-is-vignetting#pixel-vignetting)

Pixel Vignetting:

Digital cameras also suffer from pixel vignetting. Compared to optical vignetting, this type of vignetting is only applicable to image sensors. Since digital sensors are flat, their pixels are all built the same way and face the same direction. Pixels in the center of the sensor receive light rays head on at 90 degrees, while pixels in the corner receive them at a slight angle. Because of this, the sensors in the corners will receive slightly less light compared to the center, causing pixel vignetting. Unfortunately, pixel vignetting cannot be cured by stopping down the lens, since it is purely the result of the angle at which light reaches individual pixels on the digital sensor.

How can I differentiate it from regular noise? Any visual examples would be immensely appreciated.

\$\endgroup\$
5
  • 1
    \$\begingroup\$ Are you actually experiencing a problem in your photography? \$\endgroup\$
    – osullic
    Commented May 12 at 9:55
  • 2
    \$\begingroup\$ How can I differentiate it from regular noise - this is unrelated to noise. It is merely the effect that the pixels around the edges of the photo are slightly less bright compared to pixels in the middle of the photo. Generally the effect is so slight you can't see it because the image will contain objects of different colors. However you can maybe see it if you take a picture of an empty perfectly white wall that's perfectly evenly lit. \$\endgroup\$
    – slebetman
    Commented May 13 at 4:36
  • \$\begingroup\$ @slebetman Please put your answers in the answers section, even if they're short - thanks. \$\endgroup\$
    – Philip Kendall
    Commented May 13 at 8:47
  • \$\begingroup\$ It is for research purposes @osullic \$\endgroup\$
    – guialgigu
    Commented May 13 at 9:21
  • \$\begingroup\$ Thank you @slebetman, is this phenomenon of pixels slightly less bright around the edges evenly or is it randomly distributed. In the first scenario, how could it be differentiated from regular natural or optical vignetting. Under the second scenario, how could it be differentiated from noise/grain? \$\endgroup\$
    – guialgigu
    Commented May 13 at 9:21

1 Answer 1

7
\$\begingroup\$

Pixels in the center of the sensor receive light rays head on at 90 degrees, while pixels in the corner receive them at a slight angle.

This part is not true... all areas of the objective element project rays to all parts of the image. I.e. the center of the sensor is also receiving rays at an angle from all of the periphery lens areas. And some rays at the sensor periphery are also "head on." enter image description here

Unfortunately, pixel vignetting cannot be cured by stopping down the lens, since it is purely the result of the angle at which light reaches individual pixels on the digital sensor.

This part is also not accurate. It is true that some rays falling on the periphery of the image sensor have the greatest angle of incidence, but that is also mitigated by stopping down.

enter image description here

Pixel vignetting is essentially just mechanical vignetting at the pixel level due to the micro lens and recessed photosite (with front illuminated sensor).enter image description here

This is more of a design consideration... I.e. if the lens' flange opening distance is shorter/wider then the overall design will have to compensate for larger angles of incidence at the sensor. It's really only a problem if it hasn't been taken into account; or if for some reason it cannot be compensated for fully... I suppose it's also a potential issue if one were to adapt a lens that is not part of the system design.

But there are design changes that can mitigate it; backside illuminated sensors to remove pixel well depth and blocking, oriented micro lenses to redirect rays, etc. I'm not aware of any way to identify pixel level vignetting when/if it occurs as a user; but it could be measured as part of the overall QE and pixel well readout (photon-electron conversion).

\$\endgroup\$
3
  • \$\begingroup\$ Give your diagram wouldn't stopping down increase 'pixel vignetting' (which I assume to be some sort of shadow of the color filter). \$\endgroup\$
    – davolfman
    Commented May 13 at 22:44
  • \$\begingroup\$ @davolfman, No; it decreases the angle of incidence by removing the outer/wider rays. It is the refraction of the off axis rays by the micro lens that causes pixel vignetting, or rather the off axis rays to start with; not the CFA. The second diagram is quite a severe example as almost no light is entering the pixel well... I don't believe pixel vignetting is really a significant factor; focusing off axis/blocked light into the pixel well is the whole purpose of the micro lens. \$\endgroup\$ Commented May 14 at 10:12
  • 1
    \$\begingroup\$ @guialgigu, I updated my answer to address your (hidden) comment/post. It is potentially/slightly more problematic towards the periphery if this factor hasn't been fully addressed during the design of the system. \$\endgroup\$ Commented May 14 at 10:26

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.