Synopsis
		sfdipn_fb < in.rsf > out.rsf mask=mask.rsf idip=idip0.rsf xdip=xdip0.rsf both=n n4=2 niter=5 liter=20 p0=0. q0=0. order=1 nj1=1 nj2=1 drift=n verb=n pmin=-FLT_MAX pmax=+FLT_MAX qmin=-FLT_MAX qmax=+FLT_MAX eps=0.0f rect#= shift#=

The output is dimensionless (stepout in time measured in time samples).
also see sfdip

Parameters
		bool both=n [y/n] if y, compute both left and right predictions   bool drift=n [y/n] if shift filter   float eps=0.0f regularization   string idip= initial in-line dip (auxiliary input file name)   int liter=20 number of linear iterations   string mask= auxiliary input file name   int n4=2 what to compute in 3-D. 0: in-line, 1: cross-line, 2: both   int niter=5 number of iterations   int nj1=1 in-line antialiasing   int nj2=1 cross-line antialiasing   int order=1 accuracy order   float p0=0. initial in-line dip   float pmax=+FLT_MAX maximum inline dip   float pmin=-FLT_MAX minimum inline dip   float q0=0. initial cross-line dip   float qmax=+FLT_MAX maximum cross-line dip   float qmin=-FLT_MAX minimum cross-line dip   string rect#= size of the smoothing stencil in #-th dimension /auxiliary input file/   string shift#= shifting of the smoothing stencil in #-th dimension /auxiliary input file/   bool verb=n [y/n] verbosity flag   string xdip= initial cross-line dip (auxiliary input file name)