>>1020460
Another hair video. You can see my PC really struggle at the end, trying to display 20,000 curves.

I realized the calculation used to bind points was overkill. Removing 3 unnecessary sample surface nodes, increased the speed up by a lot. Reducing the cost from 420 to 250.
The joy of discovering that optimization was short lived after I made this new test hair. It's longer, and curlier. Thus requiring a lot more points. So the cost is a lot higher anyway. Plus, it required clumping, which adds even more time.

The shaping of the base mesh takes a little more work then I initially realized. But it's still overall very simple. This time, I started by forming a quad mesh around the whole scalp. Then extruded out, and added 2 loop cuts. Then used sculpting tools to shape the mesh. Using a mask on the scalp faces to prevent them from moving. the base mesh could be better sculpted, but I didn't want to fuss with it too much just for a test. When the curves are enabled, it covers up some of the wonkiness of the base mesh.
I felt the need to push up to 20,000 curves, because 15,000 wasn't enough to cover the scalp.