They may be obtained by stellating the regular convex dodecahedron and icosahedron, and differ from these in having regular pentagrammic faces or vertex figures. They can all be seen as three-dimensional analogues of the pentagram in one way or another.
These figures have pentagrams (star pentagons) as faces or vertex figures. The small and great stellated dodecahedron have nonconvex regular pentagram faces. The great dodecahedron and great icosahedron have convex polygonal faces, but pentagrammic vertex figures.
In all cases, two faces can intersect along a line that is not an edge of either face, so that part of each face passes through the interior of the figure. Such lines of intersection are not part of the polyhedral structure and are sometimes called false edges. Likewise where three such lines intersect at a point that is not a corner of any face, these points are false vertices. The images below show spheres at the true vertices, and blue rods along the true edges.
A Kepler–Poinsot polyhedron covers its circumscribed sphere more than once, with the centers of faces acting as winding points in the figures which have pentagrammic faces, and the vertices in the others. Because of this, they are not necessarily topologically equivalent to the sphere as Platonic solids are, and in particular the Euler relation
does not always hold. Schläfli held that all polyhedra must have χ = 2, and he rejected the small stellated dodecahedron and great dodecahedron as proper polyhedra. This view was never widely held.
The following images show the two dual compounds with the same edge radius. They also show that the Petrie polygons are skew. Two relationships described in the article below are also easily seen in the images: That the violet edges are the same, and that the green faces lie in the same planes.
Stellation changes pentagonal faces into pentagrams. (In this sense stellation is a unique operation, and not to be confused with the more general stellation described below.)
Greatening maintains the type of faces, shifting and resizing them into parallel planes.
The great stellated dodecahedron shares its vertices with the dodecahedron. The other three Kepler–Poinsot polyhedra share theirs with the icosahedron.The skeletons of the solids sharing vertices are topologically equivalent.(The midradius is a common measure to compare the size of different polyhedra.)
Traditionally the two star polyhedra have been defined as augmentations (or cumulations),i.e. as dodecahedron and icosahedron with pyramids added to their faces.
Kepler calls the small stellation an augmented dodecahedron (then nicknaming it hedgehog).
In his view the great stellation is related to the icosahedron as the small one is to the dodecahedron.
This is just a help to visualize the shape of these solids, and not actually a claim that the edge intersections (false vertices) are vertices.If they were, the two star polyhedra would be topologically equivalent to the pentakis dodecahedron and the triakis icosahedron.
All Kepler–Poinsot polyhedra have full icosahedral symmetry, just like their convex hulls.
The great icosahedron and its dual resemble the icosahedron and its dual in that they have faces and vertices on the 3-fold (yellow) and 5-fold (red) symmetry axes.
In the great dodecahedron and its dual all faces and vertices are on 5-fold symmetry axes (so there are no yellow elements in these images).
The following table shows the solids in pairs of duals. In the top row they are shown with pyritohedral symmetry, in the bottom row with icosahedral symmetry (to which the mentioned colors refer).
The table below shows orthographic projections from the 5-fold (red), 3-fold (yellow) and 2-fold (blue) symmetry axes.
Most, if not all, of the Kepler-Poinsot polyhedra were known of in some form or other before Kepler. A small stellated dodecahedron appears in a marble tarsia (inlay panel) on the floor of St. Mark's Basilica, Venice, Italy. It dates from the 15th century and is sometimes attributed to Paolo Uccello.
In his Perspectiva corporum regularium (Perspectives of the regular solids), a book of woodcuts published in 1568, Wenzel Jamnitzer depicts the great stellated dodecahedron and a great dodecahedron (both shown below). There is also a truncated version of the small stellated dodecahedron. It is clear from the general arrangement of the book that he regarded only the five Platonic solids as regular.
The small and great stellated dodecahedra, sometimes called the Kepler polyhedra, were first recognized as regular by Johannes Kepler around 1619. He obtained them by stellating the regular convex dodecahedron, for the first time treating it as a surface rather than a solid. He noticed that by extending the edges or faces of the convex dodecahedron until they met again, he could obtain star pentagons. Further, he recognized that these star pentagons are also regular. In this way he constructed the two stellated dodecahedra. Each has the central convex region of each face "hidden" within the interior, with only the triangular arms visible. Kepler's final step was to recognize that these polyhedra fit the definition of regularity, even though they were not convex, as the traditional Platonic solids were.
In 1809, Louis Poinsot rediscovered Kepler's figures, by assembling star pentagons around each vertex. He also assembled convex polygons around star vertices to discover two more regular stars, the great icosahedron and great dodecahedron. Some people call these two the Poinsot polyhedra. Poinsot did not know if he had discovered all the regular star polyhedra.
The following year, Arthur Cayley gave the Kepler–Poinsot polyhedra the names by which they are generally known today.
A dissection of the great dodecahedron was used for the 1980s puzzle Alexander's Star. Regular star polyhedra first appear in Renaissance art. A small stellated dodecahedron is depicted in a marble tarsia on the floor of St. Mark's Basilica, Venice, Italy, dating from ca. 1430 and sometimes attributed to Paulo Ucello.
Norwegian artist Vebjørn Sands sculpture The Kepler Star is displayed near Oslo Airport, Gardermoen. The star spans 14 meters, and consists of an icosahedron and a dodecahedron inside a great stellated dodecahedron.