Delta mush isn't a smooth di per se', it's a binding state trick. You smooth a mesh and save the delta of points from full detail to smooth in the bind pose, you then deform and smooth out the full mesh, and lastly re-apply the deltas back on the deformed smooth mesh to reinstate the high frequency detail and irregularities.
A basic implementation like that is quite simple actually, and that's why it's kinda brilliant. It's pretty easy to do with OOTB XSI actually if you use ICE to apply shapes after the envelope, or use ICE to skin things in the modelling stack. Take mesh, smooth it, shape animate from smooth to HR with shapes set to local (it will use the local reference frame of each point). Then apply your skinning, and apply a smooth deformer on top. Lastly apply the local shape through ICE to recomputed point framesets. The reference frameset is trivial, Y is the normal, X or Z is the first edge projected on the normal's plane, the last axis is the cross product of the two. You also have ICE facilities for that, I believe, and if they match the same system used by local shapes they should just work. If you instead do the skinning in ICE in the modelling stack, then you have even less work to do, as you can use local shapes OOTB (they are meant to preserve local transforms in shape with something live going on in the modelling stack). On Mon, Sep 8, 2014 at 6:28 AM, Jason S <jasonsta...@gmail.com> wrote: > What I dont understand is, how does it smooth-out the visible yanks in > the mesh, but while not also shoothing-out the nose for example? does it > use the base static mesh shape or? > > nevertheless, 100% nice! :) > > > > On 09/07/14 9:19, Miquel Campos wrote: > > 100% ICE > > > ---------------------------------------------------- > > Miquel Campos > www.miquelTD.com > > > > On Sun, Sep 7, 2014 at 10:12 PM, Sebastien Sterling < > sebastien.sterl...@gmail.com> wrote: > >> Looks cool ! is this all in Ice ? or does it require another mesh to be >> smoothed out ? >> >> >> On 7 September 2014 05:14, Miquel Campos <miquel.cam...@gmail.com> wrote: >> >>> Hello, >>> >>> I would like to share with you my Delta Mush ICE compounds (Or what I >>> guess delta mush is :P). I am pretty sure is not the best implementation of >>> Delta Mush (due my null maths skills ) but it is working kind of OK. (Very >>> slow that is true ) >>> >>> This version is using a simple average neighbours smoothing (I think >>> this is a uniform Laplacian), without any weighted average. That means that >>> can cause smoothing artifacts if the mesh have big differences between the >>> smallest and biggest polygons. This artifacts normally show up with higher >>> mush iterations. >>> >>> One little trick that I found is apply a relax before apply Delta >>> Mush. So this uniform a little the size of the polygons >>> >>> Also I found Delta mush is very impressive when you only have 1 >>> deformer x Point. is like Voodoo magic ;) >>> >>> Regarding experiments and errors, If the smoothing at origin is >>> different that the smoothing in the final, you can control how much the >>> detail is washout or exaggerate. I call it the body-builder effect. >>> >>> https://dl.dropboxusercontent.com/u/1530991/DeltaMush_ICE.rar >>> >>> I hope you like it! >>> >>> if you have any improvement or smoothing algorithm, please share it :) >>> >>> Cheers, >>> Miquel >>> >>> PS: you need to activate the delta mush in the sample scene. Check the >>> Second ICE tree in the animation stack. >>> PS2: Eddy share your OCD compounds, we want it!! ;) >>> >>> >>> ---------------------------------------------------- >>> >>> Miquel Campos >>> www.miquelTD.com >>> >>> >> > > -- Our users will know fear and cower before our software! Ship it! Ship it and let them flee like the dogs they are!
