This is shown very nicely in this particle plot in a cut in the disc for the smallest and largest stokes simulations
9/11

Next, we see if this instability also occurs globally. We run 3D global SPH simulations with different Stokes number. WInDI is triggered globally and grows VERY quickly. Here's what it looks like for Stokes=0.1
6/11

This is shown nicely in this schematic (based on values from the calculation). Top panel shows the thickness of gas (blue) and dust (red). While the gas thickness oscillates, the dust gets compressed. This effect increases with the stokes number and warp magnitude, Figs 3, 4
5/11