Where'd Shrinkwrap Go

I’ve been hearing a lot about the other developments with Blender, but I haven’t heard much about this for a while.

Anyone know what’s up?:confused:

Read this, and you’re always up to date.

http://lists.blender.org/pipermail/bf-blender-cvs/

The last message regarding shrinkwrap is:
http://lists.blender.org/pipermail/bf-blender-cvs/2008-August/015136.html

With kind regards

Mac builds are up on Graphicall :slight_smile:

I’m a LinWin guy:(.

Read this, and you’re always up to date.

http://lists.blender.org/pipermail/bf-blender-cvs/

The last message regarding shrinkwrap is:
http://lists.blender.org/pipermail/b…st/015136.html

With kind regards

That’s what I used to think:(

The mailing lists tell that Shrinkwrap can now be a constraint and the dev. is working more on his simple deform modifier. This is only in the past few days.

The only project that didn’t have an update in the Sunday Meeting was lightcuts.

Why are you sorry about that? Watch for the date of the message. And than take a look at the mailinglist. The guy is working pretty much every day at the modifier.

Now that it’s deform only, that may be applied to a multires and allows a lot of possibilities ! I am looking forward a new build on windows ; please please please someone build it !!! :slight_smile:

Cause I had no idea where to look.:o

I’ve tested this build with Shrinkwrap modfier and it’s amazing. For retopo tasks it’s very fast and does a great job sticking vertexes to other objects. Made some screenshots:

Original nose with topology that comes from subdivided (multires) cube.
http://www.uploadgeek.com/uploads456/0/retopo_final_01.png

Original + New object using “actual Retopo tool” (just flat base new object without subdivision):
http://www.uploadgeek.com/uploads456/0/retopo_final_03.png

The problem when simply subdividing the new retopologized object: it gets “deflated” by subdivision…
http://www.uploadgeek.com/uploads456/0/retopo_final_04.png

After adding “Shrinkwrap modifier” to the new object already subdivided (using options from Shrinkwrap modfier: Normal projection, Default normal and Invert normal):
http://www.uploadgeek.com/uploads456/0/retopo_final_05.png

Final object retopologized:
http://www.uploadgeek.com/uploads456/0/retopo_final_06.png

It’s amazing how simple it makes it, and it’s really fast. This build doesn’t have the new Multires modifier but I think in practice it will be more appropriate to use Multires modifier instead of Subsurf before Shrikwrap like I did here.

I think it would be great if it were possible to tell Shrinkwrap modifier to move the vertexes of the subdivided version of the new base mesh (the control vertexes from first level) instead of the new base mesh vertexes’ original positions.
In other words, modifying the base mesh in a way that, when you add Subsurf, the vertexes of the base mesh (now subdivided) would end up on the surface of the reference (sculpted) object. This way we would be able to obtain a new Base model (minimum number of faces and new clean topology) to which we could simply add Subsurf and obtain the “original sculpted” one.

It’s already a great tool, the new base nose has 108 faces (1664 when subdivided twice like in the last image), from an original (only nose) model of 2910 faces at level 7 of multires (that’s 46560 faces at level 9).

Thanks Jaguarandi!

.

Now that would be something sweet…
pardon my ignorance, but what distincts the multires modifier from the the subdivision modifier except for the ability to sculpt on it? I mean in conjunction with the shrinkwrap modifier? Bleh… what I mean is: isn’t there some functionality in these two/three modifiers that could be merged so you have less modifiers?

That is pretty amazing. I had the impression that shrinkwrap is something for cloth only, but this is something really useful.

But, in Eclectiel’s retopo example, can you apply the shrinkwrap modifier here, without applying the subsurf modifier? That I think is the difference between multires and subsurf.

Anyone know where I can get a windows build of the latest shrinkwrap? I could probably do a compile for the linux version.

Actually, can I use linux to compile a win32 version and if so how would I go about doing it?

I think the other main difference between Subsurf and Multires modifiers, besides the ablity to sculpt/modify the new vertexes, is that those new vertexes become “real”, which I guess means more memory and bigger Blend file sizes (even thought Multires modifier does magic with memory)… maybe performance issues?. An there must be technical aspects (data representation in memory??) that non-technical people like me are not aware of for a subdivision that can be edited compared to a subdivision that can not be.

I agree. At first glance seems there’s no difference since all vertexes are sticked to the other object anyway. But if you add Shrinkwrap to an object with Multires modifier, I guess you will be able to apply Shrinkwrap and then keep working on your multires mesh the way it was left by Shrinkwrap. This build doesn’t let apply Shrinkwrap if it is added after Subsurf, it actually says as an error that the modifier is not first in the stack (hope it will let in future when Multires is before it).

From a workflow point of view that sounds nice. I would not merge Shrinkwrap with the other two, but a “General Subdivision” modifier with a “Multires Button” that makes that subdivision editable, sounds good to me. Although there are always technical issues that may make not viable some things that sound simple.

.

I don’t see a reason why multires should be a modifier for this purpose (apart from perhaps memory optimization). Why not use shrinkwrap with the existing multires implementation (in the sculpt tool tab in object mode)? The “pin” value tells you what level vertices the shrinkwrap modifier affects. Then, shrinkwrap can be the first modifier in the stack, and could be applied, modifying also the base-level geometry.

That’s true, after using Multires as a modifier in another Blender build I guess I automatically searched for a modifier to replace it to test Shrinkwrap, and used Subsurf instead of using the old Multires method.

Tested it now with old multires, just to check everything is ok when applying Shrinkwrap, and it works.

I’m just wondering if there is already a method (or workaround) to use Shrinkwrap on the “control vertexes” of a subdivided model. Once you turn off Subsurf on that object you get an amazing base object, optimized and with a clean topology, that can be simply subdivided to mimic the original, and also be baked normals from the original for the small details.

.

That is exactly how multires should work. Whatever vertice movement you do at level 4, should be averaged to the lower level geometries as well, right down to level 1 (base “hull”).

For example, if you select all vertices at level 4, rotate 180 degrees in edit mode, the level 1 vertices should also get rotated. Does not matter what is moving the vertices, in this case it is an applied modifier.

Is difficult to explain only with words what I’m trying to, so I made some screenshots to see if this way the idea is more clear…

http://www.uploadgeek.com/uploads456/0/shrinkwrap_subD_final.png

With Multires it is needed to keep multires levels in order to mimic the original shape. With the other technic it is not needed to have levels of multires, the retopo Base geometry is modified, only the control vertexes because there are no other multires levels in that object. And is modified in a way that simply adding Subsurf it mimics the original surface.

We end up with a low polycount object ready to use Subsurf to become the original one.

(Finally, normals baking for the small details)

.

Thanks for the clarification and the pictures. I did understand what you are trying to do, and I agree completely.

The problem, as I see it, is that the subsurf algorithm is one-way: you take the base object and subdivide it to produce more detailed geometry. The way the modifier stack works is that the next modifier is using the output of the previous modifier. Because the subsurf is one-way, what is done on the detailed output by the next modifier has no effect on the base object (and that is why some modifiers can not be applied when they are not first in the stack).

The multires system was supposed to be a two-way algorithm: unlike the subsurf, modifications on the detailed geometry should cascade back to the base object geometry. The problem with the existing implementation is that this doesn’t always work very well. I remember seeing a workaround for this problem somewhere, that the first level added to the base object should be simple subdivision instead of catmull-clark, the next ones catmull-clark. Not sure if that would work here.

The first way I imagined it was an option in the Shrinkwrap modifier to indicate it to internally subdivide the object N times before doing it’s magic, then do the shrinkwrapping to the “control vertexes” of that subdivided version, and then (internally again) take off the subdivision to the result. Pretty much the way I did it manually in the example images.

This way Shrinkwrap could be the first modifier, and the output would be the base mesh shrinkwrapped but in a way that sets it ready to add Subsurf and automatically mimic the original model.

So if we are going to use 2 levels of Subsurf, we add in that “pre-subdivision” option of the Shrinkwrap modifier “2 levels”.

Since it would be the first modifier, one could apply it and get the final clean base mesh without modifiers.

not sure if that’s really possible…

.

So…

ummm…

>_>

Who do you have to kill to get the latest windows build around here?

:spin: