The CP 990 is put on a 1/2 inch thick plate and screwed on it by a 1/4" screw.
A blank sheet of paper.
A LASER pointer which is fixed horizontally on a support.
This assembly is glued on a long transparent ruler on which a straight line is in the vertical plane of the LASER ray.
A pencil, a protractor and a square or their functional equivalent are necessary.
The LASER power of a LASER pointer is small enough as not to damage the CCD sensor of a camera or its electronics. As a matter of fact its light flux density is much smaller than the SUN shining and burning through your front lens. However this author is not responsible for possible subsequent consequences of the repetition of this experiment by anyone else.
When the LASER pointer is aligned on the optical axis, the LASER ray illuminates the CP990 through the center of the lens. The Aspect Angle is null.
The CCD is then completely saturated by the red LASER and this can be observed on the LCD on the camera itself. A picture shows that effect.
Turning the LASER pointer (even slightly) away from this position shall desaturate the CCD. The picture shown on the LCD is normal. The LASER is shown as a very small red dot.
Move the LASER pointer in front of, but offset from the center Optical Axis of the Fisheye. Then, by slowly rotating the pointer assembly around itself, a specific new position shall produces again saturation of the camera CCD.
Record (protractor) the corresponding Aspect Angle ß between the Optical Axis projection and the line on the ruler by reproduction of the same line on the paper sheet.
The intersection of this line with the Optical Axis is the nodal point for this ß aspect angle.
Repeat this step for various angles on each side of the optical axis from ß0 to ß95deg which is the limit of the lens field-of- view (FOV).
see the figure below
Actually the nodal point location depends on the aspect angle ( ß ) from the optical axis of the lens.
There is an INFINITY of nodal points. All are located at a distance (D) that goes from 17.5 mm from the lens front to 4.5 mm from the same surface. That is a 13 mm range along the optical axis (i.e. about two focal length !)
Suppose that two objects (one near the camera and another farther away) are aligned on the main optical axis (i.e.ß=0deg)
When you rotate the camera around an axis which contains a point which belongs to the 13 mm range, you shall have two (and only two) other different positions where the pair of points looks again aligned (i.e. the closest hiding the farthest one). The first position give a ß angle and the other is opposite and symmetrical (-ß)
As the lens is geometrically symmetrical AROUND the optical axis, the above property is valid for the half-conical angle ß in the global space (in three dimensions).