7
\$\begingroup\$

Vignetting is usually assigned to lens characteristics, or caused by light-blocking objects in front of a lens, like lens-hood or a filter.

Is there vignetting (or vignetting-like effect) caused by sensor too? When light enters camera body thru a lens, it naturally is not falling on sensor "head-on" but rather in an angle. At ever deeper angle the nearer the sensor edge the light is falling on.

A normal SLR body has quite a lot of room between lens rear element and sensor, so I'd guess sensor-originated vignetting is not a big problem after all. But Sony has now made a Full-frame NEX camera. I wonder how large a rear element in lenses they need to design for this camera, where lenses are mounted so near the sensor. How does a modern image sensor fight vignetting caused by the changing (center vs edge) angle of light rays?

\$\endgroup\$
5
  • \$\begingroup\$ Another question about lens mount to sensor distance. \$\endgroup\$ Commented Oct 15, 2013 at 18:36
  • 3
    \$\begingroup\$ A way of avoiding this problem is to make the lenses image-space telecentric, sometimes called rear telecentric. That means that all the light rays hit the sensor at right angles. Four Thirds and Micro Four Thirds have telecentricity as part of the specification, but i'm not sure how strict it is. I can certainly use non-telecentric legacy lenses on my µFT camera without noticeable problems. \$\endgroup\$ Commented Oct 15, 2013 at 19:33
  • \$\begingroup\$ @TomAnderson the µFT sensor is small compared to the flange distance, the problems are much worse when using APS-C or full frame sensors with the same short flange distance. \$\endgroup\$
    – Matt Grum
    Commented Oct 15, 2013 at 19:46
  • 1
    \$\begingroup\$ @MattGrum: µFT's flange distance is a bit bigger than those of the APS-C mirrorless systems, but quite a bit smaller than any SLR. The ratio of sensor diagonal to flange distance for µFT is 22.5 mm / 19.25 mm = 1.17, whereas for Canon EF with a full-frame sensor it's 43.3 mm / 44.00 mm = 0.98, and with an APS-C sensor it's 26.7 mm / 44.00 mm = 0.61 (although the back end of an EF-S lens can go deeper in than the mount, right?). So the sensor is actually bigger relative to the flange distance than in an SLR. \$\endgroup\$ Commented Oct 15, 2013 at 20:23
  • \$\begingroup\$ But i suppose the thing with a legacy lens on an µFT camera is that (a) the adaptor shunts it forward to its normal flange distance, and (b) the 2x crop factor means the more steeply inclined rays at the edge of the lens's image circle are being thrown away anyway. \$\endgroup\$ Commented Oct 15, 2013 at 20:24

2 Answers 2

10
\$\begingroup\$

Sensor based vignetting is a real problem for cameras with a large sensor and short flange distance (distance from mount to sensor). This is simply due to the oblique angles of incidence of light hitting the edges of the sensor, which isn't really flat like a piece of film, but contains a stack of objects starting with UV/IR/AA filters, followed by microlenses, colour filters, then many layers of silicon with the photosensitive area at the bottom of a deep well!

Near telecentric lens designs with large exit pupils largely solve the problem, but at the expense of making the lenses larger and more expensive.

Several things can be done to address the problem. Minimising the thickness of the filter stack certainly helps, though Leica attempted this with the M8 but made the IR filter too thin making the camera very sensitive to IR light, causing strange colours in the image (they even had to give out free IR screw on filters to angry customers).

Doing away with the anti-aliasing filter is another option that Leica and Sony (with the A7r) have pursued. Another option is offset microlenses (Leica M9) or elliptical microlenses (Leica M240), both of which attempt to aim the oblique light rays more effectively toward the photosites. The problem with this is that it compromises performance with longer focal length lenses.

The best solution is to get the photosensitive area of the silicon as close to the top as possible. One way is to thin the back of the wafer until it becomes transparent and then flip the sensor around so the readout circuitry is behind the photosite. These so called "back side illuminated" sensors are in existence, but only very recently in anything but very small cell phone cameras.

Finally organic sensors have been touted as the solution to ray angle issues, these have an organic layer that is photosensitive at the very top with circuity underneath. They are a way of production, but promise to eventually deliver the utopia of tiny wide angle lenses that ought to have been the result of doing away with the DSLR mirror!

\$\endgroup\$
3
  • \$\begingroup\$ As an interesting piece of trivia, our own sensors - our retinas - also have an inverted structure, with the light sensitive cells hiding behind support machinery. en.wikipedia.org/wiki/… \$\endgroup\$ Commented Oct 15, 2013 at 21:15
  • \$\begingroup\$ Sony A7R has offset microlens sensor design too, but Sony A7 has not. \$\endgroup\$ Commented Oct 21, 2013 at 19:13
  • \$\begingroup\$ And maybe curved sensors? \$\endgroup\$
    – Rolf
    Commented Apr 8, 2019 at 1:41
4
\$\begingroup\$

In all cameras lens design affects lens vignetting. The lens has an 'image circle' which is designed to match the size of the film/sensor (which is why on FX cameras DX lenses vignette a lot). For digital cameras there is an additional source of vignetting called pixel vignetting.

In short, the individual pixels of the sensor physically sit at the bottom of an opaque well (the well is made of structural components and sensor electronics). This well acts as a collimator and reduces the intensity of any light that comes at an angle to the sensor. At the middle of the sensor the light is coming in parallel to the well and strikes the pixel full-on, generating the largest signal. At the edges of the sensor the light strikes at an angle and is largely cutoff by the well, resulting in lower sensor signal.

Camera manufacturers calibrate the vignetting of the sensor and correct for this in software. A detailed explanation is available in this research article. Some vignetting is also countered by having microlenses at the top of the wells to focus the light down the well.

\$\endgroup\$
2
  • \$\begingroup\$ But how do they overcome this issue? Is vignetting really not a lens characteristic at all, but in the sensor design? \$\endgroup\$ Commented Oct 15, 2013 at 19:03
  • \$\begingroup\$ It is both. Vignetting occurs due to the lens. The lens has an 'image circle' which is designed to match the size of the film/sensor (which is why on FX cameras DX lenses vignette a lot). For digital cameras there is additional vignetting occurring due to the senssor. The sensor vignetting occurs because of the geometry of the sensor. Apparently camera manufacturers know the characteristic of the sensor vignetting and correct for this using software. What I don't know is if the raw files will have the correction built in or not. That's actually a good second question to put up! \$\endgroup\$ Commented Oct 15, 2013 at 19:07

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.