Spectre
Experiments with VDBs/SDFs for curve/point manipulation.
Process
Curve/point advection via distance from SDF
I’ve been learning how to use volumes/SDFs to attract nearby points. I found that it worked extremely well for shrink wrapping randomly scattered curves around a head mesh, looking almost like a fibreglass surface. Perhaps an improvement would be to only advect the points greater than the SDF boundary, which would keep the interior of the mesh dense with points/curves, rather than hollow.
Cheap wire sag setup
For simple wire sag effects, I usually use the curveu attribute of the resampled curves to create a bend/falloff attribute, pow(@curveu*(1-@curveu),(1/falloff)) usually does the trick.
However this time I wanted to only sag the wires in certain places, e.g. on the more sparse areas of curve distrubution, not the dense parts which define the facial features. By doing a pretty heavy point smooth, then finding the distance from that smoothed @P and the original non-smoothed @P, I could create an attribute map with which to multiply my curveu gradient by.
Attribute map from hell. Smoothness map to define which areas to sag.
Wire sag using curveu*smooth attrib, subtracted from @P.y.
Final shot, with wire sag easing in over time.
As a cherry on top, you could probably run the curves through a noise modifier prior to the @P.y subtraction, then apply the inverse displacement to put it back in a similar place afterwards. The melting should then flow in some way rather than just directly down.