Easiest way to smooth a strand is to do the following for each strand point:
p(n) = A x p(n) + (1-A) x 1/2 x ( p(n-1) + p(n+1) ) Where: p(n) = Current strand point position p(n-1) = Previous strand point position p(n+1) = Next strand point position A = Smoothing factor (from 0 to 1) For best effect, do this for a number of iterations keeping A as small as you can. You can get better results averaging over more neighbouring points in the strand and weighting them using an appropriate filter (Gaussian, etc.) A > On 2 Jan 2015, at 14:27, Olivier Colchen <[email protected]> wrote: > > Hello, > > To be honest, I think you're both right : it's probably not worth the time, > especially considering I don't even have a month to setup everything properly > in each scene and then render it. And only a handful of frames would actually > benefit from that. I barely tried the dynamic framework solution, as, as you > mention Stefan, it seems to only really work on fixed-length strands, as well > as setting up some flex/stretch I don't really want. And devling deeper into > the system looked time-consuming. > > > However I believe I was quite near the result I wanted, going back to the > simple Closest Location to strandposition compound. Maybe I could find the > middle solution and smooth the strands whenever they leave the volume to fix > the little dents appearing here and there. I have no idea how to smooth a > strand, though... > > Olivier > > 2015-01-02 15:14 GMT+01:00 [email protected] <[email protected]>: >> >> >> > On 2 Jan 2015, at 12:39, Stefan Kubicek <[email protected] >> > <mailto:[email protected]> > wrote: >> > >> > >> > PS: If the runners trails are meant as a graphical element rather than >> > an >> > actual physical element in your picture I doubt I will look nice when it >> > collides with the runner. >> > >> >> >> >> Agreed. I doubt it would give you a look that you want and probably isn’t >> worth >> spending too much time trying to make a robust system to make it work. Do a >> few >> quick tests first to see if it’s going to take you in the right direction. >> >> As a quick alternative, why not try setting the StrandColor to black for any >> strand points that fall inside your collision geometry, and then incorporate >> that as a multiply operation into your shader? That'll hide any parts of the >> strand that go inside your character. >> >> A >> >> >> >> On 02 January 2015 at 12:39 Stefan Kubicek <[email protected]> wrote: >> >> >> > Hi Olivier, >> > >> > I think there are example scenes for strand collision that ship with XSI. >> > No >> > friction between strands and geo is modeled as far as I can remember, but >> > it >> > might get you started. >> > >> > Then there is also the strand collision framework for more accurate >> > collisions, but I don't know how it reacts to changing point counts and >> > strand >> > length: http://softimage.tv/strand-collision-framework/ >> > >> > PS: If the runners trails are meant as a graphical element rather than an >> > actual physical element in your picture I doubt I will look nice when it >> > collides with the runner. Maybe you could solve this in comp and get around >> > explaining collisions entirely? >> > Here is a nice example in a recent Glassworks spot that might be similar >> > to >> > what you to try to achieve: >> > http://softimage.tv/lycra-moves-you/ >> > >> > Good luck, >> > >> > Stefan >> > >> > >> > >> > >> > > > Hello, long time reader, first time poster here! >> > > >> > > >> > > So I have a bit of a problem on my hands. >> > > >> > > >> > > We are doing a very simple concept, a sportsman whose back is leaving >> > > strand trails as he goes. This is child's play so far. >> > > My issue is that we would like the trails to interact with the mesh >> > > when, >> > > for various reasons, he cuts through them again. >> > > IE : at some point he falls to the ground and stands back up. The >> > > unwanted >> > > behaviour is that the strands generated from the back of his head will >> > > simply go through his head as he rises up, resulting in a somehow >> > > vertical >> > > strands column masking his face and fore-body. The desired behaviour is >> > > that >> > > the strands will flow around his cranium, maybe along his back, >> > > effectively >> > > leaving at least his face apparent. >> > > >> > > >> > > Animation is done in Maya and transfered over via geocache. >> > > >> > > I tried to operate with tools like 'closest location' or 'get location >> > > by >> > > raycast' applied to the strandpositions when inside the volume to >> > > simulate >> > > collision, but these were all lacking. The raycast solution may be >> > > promising, but I don't know how to set the correct direction vector. And >> > > sometimes the animation I'm working with may "skip" a strandposition, >> > > effectively not triggering the 'inside volume' condition. >> > > >> > > >> > > So yeah basically I don't know how to do that in a visually pleasing >> > > way. >> > > If anyone had some pointers, I'd be very grateful ! >> > > >> > > Thanks, >> > > Olivier >> > > >> > > >> > > >> > > > >> > >> > -- >> > >> > ----------------------------------------------------- >> > Stefan Kubicek <a href="mailto:"[email protected]"; >> > <[email protected]>">[email protected] >> > ----------------------------------------------------- >> > Alfred Feierfeilstraße 3 >> > A-2380 Perchtoldsdorf bei Wien >> > Phone: +43 (0) 699 12614231 >> > www.keyvis.at <http://www.keyvis.at> >> > This email and its attachments are >> > confidential and for the recipient only >> > >> > >> > >
