--transform a,b,c,d,e,f,g,h,i
              Specifies  a  transformation matrix to apply on the output. Automatically a
              bilinear filter is selected.  The mathematical form corresponds to:
                     a b c
                     d e f
                     g h i
              The transformation is based on homogeneous coordinates. The  matrix  multi‐
              plied  by  the  coordinate vector of a pixel of the output gives the trans‐
              formed coordinate vector of a pixel in the graphic buffer.  More precisely,
              the  vector (x y) of the output pixel is extended to 3 values (x y w), with
              1 as the w coordinate and multiplied against the matrix. The  final  device
              coordinates  of  the pixel are then calculated with the so-called homogenic
              division by the transformed w coordinate.  In other words, the device coor‐
              dinates (x' y') of the transformed pixel are:
                     x' = (ax + by + c) / w'   and
                     y' = (dx + ey + f) / w'   ,
                     with  w' = (gx + hy + i)  .
              Typically,  a and e corresponds to the scaling on the X and Y axes, c and f
              corresponds to the translation on those axes, and g, h, and i  are  respec‐
              tively  0,  0  and  1.  The matrix can also be used to express more complex
              transformations such as keystone correction, or rotation.  For  a  rotation
              of an angle T, this formula can be used:
                     cos T  -sin T   0
                     sin T   cos T   0
                      0       0      1
              As a special argument, instead of passing a matrix, one can pass the string
              none, in which case the default values are used (a unit matrix without fil‐
              ter).