The stereographic cylindrical projection designed by C.Braun can be visualized geometrically, but it is slightly more complicated than the previous azimuthal orthographic example. Instead of a flat projection plane directly yielding the projected map, here the projection surface is a cylindrical sheet tightly rolled against the Equator. Every meridian is drawn on this tube by light rays emanating from an equatorial point on the meridian directly opposite. The tangent tube is then cut along an arbitrary meridian and unrolled.
There is a similar procedure for creating some cylindrical projections, although not all of them employ such a simple model. In fact, well-known projections like Mercator's and the equidistant cylindrical are defined by arbitrary constraints, not a perspective process.
Equations for direct mapping of the normal aspect are straightforward. Consider projecting point on Earth onto point on the map. First, like in all cylindrical designs, meridians in this aspect are projected as straight vertical lines with spacing directly proportional to longitude only, therefore .
The parallel determined by is projected as a straight horizontal line, whose -coordinate can be derived from the diagram by a simple proportion:
Making and using the trigonometric identities
for the range , the ordinates simplify to
Again, inverse mapping equations can be easily obtained.
On Braun's projection, the cylindrical surface is tangent and the Equator is the standard parallel. The derivation above can be immediately generalized to a different standard parallel just by reducing the cylinder's radius to . With a secant surface, the first proportion becomes
Thus the equations for stereographic cylindrical projections like Gall's ( = 45°) and the BSAM cylindrical ( = 30°) are
Finally, it's possible shifting the light source from a point on the Equator, i.e., at a distance from the polar axis, to a distance . The equations, without the trigonometric substitution above, become
Two particular cases are Braun's pseudo-Mercator projection with and the central cylindrical projection, with (both use = 0°).
Across all cylindrical projections, the only essential difference is the vertical scale, which in the equatorial aspect translates to the spacing between parallels. Most such projections, like Braun's and Gall's stereographic, are designed to increase the spacing towards the poles. Why?
Consider a series of graticule "cells" of uniform "width" and "height" (say 10°); on Earth, they are approximately square at the Equator but with growing latitudes they progressively narrow down, becoming very thin spherical triangles at the poles. By definition, cylindrical projections cannot narrow their cells, only stretch them vertically as an approximation. In particular, in Mercator's the stretching is calculated in order to exactly preserve local angles; this requires the poles to be projected into infinity.
Of course, vertical stretching exacerbates the exaggeration of area typical of high latitudes in cylindrical maps. In contrast, Lambert's equal-area cylindrical projection and its variants preserve areas by compressing instead of stretching latitudes, but shapes suffer accordingly. This kind of compromise is a recurrent theme of map projections.