Note that this reference documentation is identical to the help that is displayed in MATLAB when you type “help ft_warp_apply”.

  FT_WARP_APPLY performs a 3D linear or nonlinear transformation on the input
  coordinates, similar to those in AIR 3.08. You can find technical
  documentation on warping in general at http://bishopw.loni.ucla.edu/AIR3
 
  Use as
    [warped] = ft_warp_apply(M, input, method, tol)
  where
    M        vector or matrix with warping parameters
    input    Nx3 matrix with coordinates
    warped   Nx3 matrix with coordinates
    method   string describing the warping method
    tol      (optional) value determining the numerical precision of the
              output, to deal with numerical round off imprecisions due to
              the warping
 
  The methods 'nonlin0', 'nonlin2' ... 'nonlin5' specify a
  polynomial transformation. The size of the transformation matrix
  depends on the order of the warp
    zeroth order :  1 parameter  per coordinate (translation)
    first  order :  4 parameters per coordinate (total 12, affine)
    second order : 10 parameters per coordinate
    third  order : 20 parameters per coordinate
    fourth order : 35 parameters per coordinate
    fifth  order : 56 parameters per coordinate (total 168)
  The size of M should be 3xP, where P is the number of parameters
  per coordinate. Alternatively, you can specify the method to be
  'nonlinear', where the order will be determined from the size of
  the matrix M.
 
  If the method 'homogeneous' is selected, the input matrix M should be
  a 4x4 homogenous transformation matrix.
 
  If the method 'sn2individual' or 'individual2sn' is selected, the input
  M should be a structure based on nonlinear (warping) normalisation parameters
  created by SPM8 for alignment between an individual structural MRI and the
  template MNI brain.  These options call private functions of the same name.
  M will have subfields like this:
      Affine: [4x4 double]
          Tr: [4-D double]
          VF: [1x1 struct]
          VG: [1x1 struct]
       flags: [1x1 struct]
 
  If any other method is selected, it is assumed that it specifies the name of an
  auxiliary function that will, when given the input parameter vector M, return an
  4x4 homogenous transformation matrix. Supplied functions are 'translate', 'rotate',
  'scale', 'rigidbody', 'globalrescale', 'traditional', 'affine', 'perspective',
  'quaternion'.
 
  See also FT_WARP_OPTIM, FT_WARP_ERROR