Graham

You got me thinking about this, but I am sure it is the school textbooks of the 19th century. Straight out of my head and it could be very wrong, I intuit the following.

A hole through a thick sheet will cause a small exit pupil and huge vignetting as the light has to pass through a tunnel, and the light rays going straight through will be OK but those at an angle will hit the sides. So, the ideal pinhole should have just atomic thickness at the edges and you should chamfer the edges to give the angle of entry you want.

The next problem is the diameter of the hole. If you want to optimise resolution you have to balance two effects; the size of the hole and diffraction. The sensor will not be able to resolve objects closer than the diameter of the hole so you want the hole to be small. But, the smaller the hole, the greater of Airy disk of diffraction. My intuition is that optimal resolution at the centre of the sensor is when the diameter of the hole is equal to the diameter of the Airy disk. (And this resolution is at the centre but gets worse to the edges.) If d is the diameter of the hole, and the wavelength of the light is L, and the distance to the sensor is F, using standard formulae, 1.22LF/d = d. Using a flange-sensor distance of 44mm for the EF mount and 400nm light, this gives d = 0.146mm, which leads to an f-number of 300.

You might wish to lower resolution by having a wider hole to reduce exposure time.

A 0.146mm hole with the effects of the size of hole on the two aspects of resolution, would give an effective pixel size of about 0.2mm. So, a 24x36mm sensor would have an effective resolution of about 22kpx, not too dissimilar from the first Casio digicam.

Please correct me if I am wrong.