Thanks for the reply, chrisjcrocker.
I feel just fine following your lead. Any problem that is not solved, is something that will come back. So, I rather dig deeper now. Of course, I would love to have a “Technicals Director’s Manual”, (which does not exist, AFAIK) and which would provide the precise explanation for anything. […day-dreaming now].
However, things are changing constantly. What was working yesterday might not do so anymore today, or is even better by now. Today, I went from installing Unity 2018.1, to installing a demo of TurbulenceFD, to Hair, to FBX vs Alembic export, to a Mesh-Deformer vs Skin Deformer, and tomorrow we have perhaps a complete new mix So, just to excuse myself that I sometimes have not the final answer.
I guess you know all of that, but since we are in a forum, here are some average thoughts. Dynamics is of course just a simulation, or in other words a simplification of the real thing. Think about a light ray or a single photon, it bounces many millions of times and all without any math, just bouncing and done. To get this handled in a computer would take very long, especially for a few million photons … Hence the algorithm tries to cut corners, as in setting up a threshold, e.g., anything lower than x% of strength will be killed and not longer bother the cpu.
Similar: the dynamics with hairs. If we really would calculate 200K hairs, how they interact, that would take forever. Hence the few guides, which makes a huge difference. But also here, the Hair is not calculated, the points are. Then we have colliders, and their surfaces are more based on points and edges than the polygon itself – AFAIK. This helps to reduce time, but also limits the precision. In reality there is no sampling, there is a flow of continuesly forces, not only just 100 times per second. Another time saver (e.g., 100 times). Well, all of that works like a fisherman’s net, if it is too wide we miss, if it is too small, we might catch a lot of what we don’t like (in our case: wasting time). On top of that we like to have a responsive system.
With that all being said, there is for each set up a sweet spot. What works for one, might miss the point for another set up. Exactly this makes me feel uncertain at times, so I have to explore it and use any information I get about the project to find that sweet spot. Even if I kill gravity along the way a little bit. Which pulls constantly. Or use the Rest values to mix the Set as Dynamic state into the result.
That you get better results now is all I’m after, besides that you hopefully feel as well happy to set things up again and enjoy your results.
My best wishes for your project.