Would not work because the start/stop transient produces about a foot of useless filament.

I knew there had to be a catch somewhere - didn't think about start/stop problems though, but it seems obvious now :-).

Suppose your filastruder could be convinced to deposit filament in a ring inside a cylinder (possibly by something a little more complicated than a filawinder).

Then suppose you pulled filament out a small hole in the bottom of the cylinder from the inside of the ring through a bowden tube to your printer.

When the weight of the cylinder dropped enough, it would turn on the filastruder and replenish the supply of filament.

Could it work, or is it more likely to get the filament twisted up inside the cylinder?