Quote from: Edwin Herdman on July 10, 2012, 06:42:28 PM
We're not talking about vignetting at the edges of an image. We're talking about photosite wells blocking photons arriving at oblique angles, "per photosite vignetting" or shadowing, if you like.
Up until I wrote that post I thought that the cutoff was at f/2, but apparently it's actually f/2.8. That's rather disappointing. It probably changes from camera to camera slightly, depending on the sensor and photosite design (at the very least, it could change or someday be "fixed" in a new design).
I think what Edwin is referring to are these findings from DxO and Luminous Landscapes back in 2010.
I had never hear of this before, and was sceptical at first, however it does make sense. However, people are clearly overreacting. If you take the time to examine the graphs on the DxO-site, you will see, that by going from f4 to f2.8 on the 7D, you up the gain with about 0.04 eV thus gaining a total of 0.96 eV of light gathering ability instead of 1 eV as expected. This is marginal and of no consequence. Going from f2.8 to f2.0 up the gain by 0.05 eV gaining 0.95 eV in light gathering power, again very little effect. Going from f2.0 to f1.4 ups the gain with 0.25 eV thus gaining only 0.75 eV of light gathering instead of 1 eV. This is worse, but you still get a three fourths of 'what you pay for', not, as indicated, nothing. In going from f1.4 to f1.2 the gain is upped by 0.2 eV giving only 0.13 eV of light gathering power instead of the expected 0.33 (a third of a stop) so this is clearly getting worse.
In total, going from f4 to f1.2 (3.33 stops) you get about 0.6 eV of sensor gain and thus, only get 2.73 eV of light gathering instead of the full 3.33 eV.
Of course the manufactures are still 'cheating', but it is not like you get nothing from using faster glass, you just get (a little) less than you thought you would.
Hope this clears things up a bit