If the sensor was a source of vigentting in the image then it would still be apparent even when stopped down to f/11.
Not true: at larger apertures, the broader light cone means that some of the light reaching the corners of the frame strike the sensor at a more off-perpendicular angle than the principle ray, potentially producing more microlens vignetting than at small apertures. About an extra 10║ at f/2.8 and an extra 21║ at f/1.4.
Also, shifted lenses are mor prone to this than any normal lens, as they increase the off-pependicular angles. That is probably why most sensors for MF and LF backs continue to omit microlenses: to accomodate camera/lens motions like shift.
The trouble in detecting this is that there are three factors producing fall off at the corners: true vignetting, microlens vignetting, and "geometric fall off" with wide angle lenses.
- True vignetting is partial shading of the frame corners by the lens barrel, filters etc.
- Microlens vignetting is my name for the reduced sensitivity of sensors with microlenses to off-perpendicular light
- geometric fall-off is the fact that with almost any wide angle lens design and any sensor including film, light travels significantly further from the exit pupil to the corners of a wide angle frame than to the center, diminishing its intensity by the inverse square law. Near telecentric lens designs have no effect on this.
It might be that digital scrutiny has enabled detection of geometric fall-off that mostly went unnoticed with film. The Olympus E system includes software options for correcting "corner light fall-off" which is probably aimed at this effect, not microlens vignetting.