Currently i did some tests and it shows that the behavior of projection and and nearest surface point is way different. Projection also makes the vertices jumping. I tested it with svn rev34989 and appended an very simple example. Just use the shapekey-slider (Key 1) of the red inner object (something like a muscle) and you will see that the skin is snapping. First nothing, then a jump, followed by a smooth further transition. That is unacceptable for muscles and in many complex cases even worse.
Example file: http://www.pasteall.org/blend/5290 Am 19.02.2011 10:16, schrieb Campbell Barton: > Found this can be done already (sort of), > Project negative then cull front or back. > > There was a bug where rotating the object made culling fail, fixed r34986. > > This works ok with a sphere/cube and sphere/plane so in real usage > there may be some other options needed. > If this isn't working well I'd be interested to see an example blend file. > > On Fri, Feb 18, 2011 at 8:02 PM, Daniel Salazar - 3Developer.com > <[email protected]> wrote: >> Thumbs up for suggestion. For group behavior lets just wait for nodal >> modifiers, this will be a no brainer with a forgroup loop node >> >> cheers >> >> Daniel Salazar >> >> >> 2011/2/18<[email protected]>: >>> Hi >>> >>> May be my recent released relaxation patch can help >>> http://www.box.net/shared/ddhtfnsq75 >>> >>> >>>> HI guys! >>>> >>>> As I see it, shrinkwrap is lacking of two things that would make it even >>>> better. >>>> >>>> 1_ A Neighboring vertexes smoothing factor. This would be done by applying >>>> a kind of falloff to the affected surface like in these examples: >>>> >>>> http://chrisevans3d.com/tutorials/maya_muscle.htm >>>> >>>> 2_ An algorithm that determines that if a vertex movement goes out of a >>>> certain distance range, those vertexes should be smooth out or its >>>> movement averaged with the movement of the neighboring vertexes. This >>>> feature could avoid vertex artifacts with an articulated object, specially >>>> in projection mode. >>>> >>>> Maybe this is too much to ask from Campbell's weekend quick fix, but you >>>> never know when Jaguarandi is reading :p >>>> >>>> >>>>> Date: Fri, 18 Feb 2011 14:26:55 +0100 >>>>> From: [email protected] >>>>> To: [email protected] >>>>> Subject: Re: [Bf-committers] Inverse Shrinkwrap? (only "keep above >>>>> surface") >>>>> >>>>> Using a group of objects could be useful, but if it's not done for >>>>> multiple objects, you still can just apply multiple shrinkwraps in "keep >>>>> above only" mode to achieve more or less the same result. In many cases >>>>> it would be just one "collision" object and fine. Of course groups of >>>>> objects would move the vertices more correctly. But as you said: One >>>>> step at a time. The first one alone would already be great. >>>>> >>>>> Am 18.02.2011 14:14, schrieb Campbell Barton: >>>>>> You're right that this is simple to add and agree it seems very >>>>> useful, >>>>>> quick way to reducing intersections without physics errors. >>>>>> The main downside I can see, as the verts slide over large variance it >>>>>> could give visible popping in some cases, though this is no difference >>>>>> then shrink-wrap for other uses. >>>>>> >>>>>> Though it makes me think for this feature it would be even more useful >>>>>> in production is if the modifier could select a group of objects to >>>>>> 'not pass through', rather then a single one. >>>>>> >>>>>> But one thing at a time, this seems reasonable and if theres no >>>>>> objections I can add this over the weekend since I made some changes >>>>>> to this area recently. >>>>>> >>>>>> On Fri, Feb 18, 2011 at 12:43 PM, Tobias Oelgarte >>>>>> <[email protected]> wrote: >>>>>>> Shrinkwrap is designed to force vertices onto the surface of another >>>>>>> object. In Nearest Surfacepoint mode it moves vertices either to the >>>>>>> surface or optionally let it also move outward until they are on the >>>>>>> surface. So you can work with essential two options. But wouldn't it >>>>>>> make more sense to include also the possibility that vertices can >>>>> also >>>>>>> only be moved up onto the surface and and not moved toward it? >>>>>>> >>>>>>> It would allow you to throw a ball against a window, making the >>>>> vertices >>>>>>> stop that would pass trough it. There are many possible usages i >>>>> could >>>>>>> think of: Bubbles in a glass. Shoes that not sink slightly into the >>>>>>> ground. Body parts that "collide" with each other, and many more. >>>>>>> >>>>>>> Since the algorithm seams to handle this case already -- knowing if a >>>>>>> vertex is inside or outside the target, for "Keep above surface" -- >>>>> it >>>>>>> should be just an additional option. Am i right? >>>>>>> >>>>>>> Cases (now): >>>>>>> * Shrink to Surface >>>>>>> * Shrink to Surface and move outward for "keep above" >>>>>>> >>>>>>> Cases (desired): >>>>>>> * Shrink to Surface >>>>>>> * Only "keep above" >>>>>>> * Shrink to Surface and move outward for "keep above" >>>>>>> >>>>>>> Hope this is a little inspiration. It would make shrinkwrap much more >>>>>>> awesome. >>>>>>> >>>>>>> Greetings from >>>>>>> Tobias Oelgarte >>>>>>> _______________________________________________ >>>>>>> Bf-committers mailing list >>>>>>> [email protected] >>>>>>> http://lists.blender.org/mailman/listinfo/bf-committers >>>>>>> >>>>>> >>>>> _______________________________________________ >>>>> Bf-committers mailing list >>>>> [email protected] >>>>> http://lists.blender.org/mailman/listinfo/bf-committers >>>> _______________________________________________ >>>> Bf-committers mailing list >>>> [email protected] >>>> http://lists.blender.org/mailman/listinfo/bf-committers >>>> >>> >>> _______________________________________________ >>> Bf-committers mailing list >>> [email protected] >>> http://lists.blender.org/mailman/listinfo/bf-committers >>> >> _______________________________________________ >> Bf-committers mailing list >> [email protected] >> http://lists.blender.org/mailman/listinfo/bf-committers >> > > _______________________________________________ Bf-committers mailing list [email protected] http://lists.blender.org/mailman/listinfo/bf-committers
