On Sep 9, 2005, at 2:27 PM, J. C. O'Connell wrote:
I did a quick sketch to clarify what I said: http://www.jcoconnell.com/temp/rearanglediagram.jpg
Your sketch is misleading: it exaggerates the relative sizing of the sensor target compared to the lens and also does not indicate where the nodal point is. In a typical Cooke triplet, it's the distance from the nodal point to the imaging plane that determines the deviation from the orthogonal as you approach the edge of the film/ sensor format, not the distance between the rear element and the film/ sensor.
The point of having a lens designed for a digital sensor that has its rearmost element very close to the sensor plane is that the rearmost elements of the lens performs correction designed to orient the light path from the nodal point (placed sufficiently far forward in the lens) such that the ray trace to the photosite plane is orthogonal, not that you'd place the nodal point further rearwards in the lens.
This is quite similar to what a condenser enlarger head does: it positions a collimating lens group very close to the film plane in order to make the light pass evenly through all points of the negative, right to the corners, and oriented orthogonally through the film so that a flat field imaging objective (the enlarging lens) will exhibit very little light falloff at corners and edges.
A large diameter element at the rear of a lens designed for the digital sensor helps in promoting this even illumination of the entire sensor area. Placing this rear lens group close to the sensor, relatively distant from the nodal point, allows the strength of the elements to be lower and thus promotes less distortion from the correction.
Godfrey
