On Mon, Apr 11, 2016 at 6:19 AM, Christopher Crouzet < christopher.crou...@gmail.com> wrote:
> The 6/4 result is to be expected and is not related to a rounding issue > but simply to the nature of randomness—setting a ratio of 0.5 doesn't > guarantee a 50-50% split, it only means that the distribution will tend > towards this goal, especially when the number of samples will be high > enough (and 10 samples definitely isn't much at all). > > Yes! ICE factory nodes have inherent problems with small sample sets (precise numbers of particles). "Generate Sample Set," for example, which lurks within many factory compounds, but cannot be edited, has this tendency. I think this is one reason you find so many people who do deterministic particle animations (e.g. non-simulated ICE trees for strand animations like, say, LKLightning) end up building their own "emitters" and other compounds from really low-level components or in C++. I ran into this recently when trying to use factory nodes to build a very simplified simulation framework that operated on very small populations of particles that had to emit at precise locations at precise intervals. Every so often, I'd get some "missing" particles and some doubled ones. I think this actually happens in simulations all the time, but we don't care since we're generally looking for large numbers of particles and statistically-accurate aggregate behavior rather than precision.
------ Softimage Mailing List. To unsubscribe, send a mail to softimage-requ...@listproc.autodesk.com with "unsubscribe" in the subject, and reply to confirm.