PDA

View Full Version : Reverse Subdivision Deformation



probiner
08-08-2010, 07:07 AM
Yesterday i suggested in 3D-Coat foruns that there would be a way to reverse the SubD deformation in a retopo mesh since this sticks to the sculpt and when it is Subdivided in LW it does not match the Sculpt in 3DC. the solution for now is to make dense meshes so that the deformation, its virtually invisible.

http://i153.photobucket.com/albums/s202/animatics/Lightwave/Subdivision-Distortion.png

But, it seems that accoding to Andrew's twitter, the next 3DC version will adress this issue.
I have add new very important option to export setting "Perform coarsing". It is inverse to subdivision.
4:29 AM Aug 7th via TweetDeck

So my question is, wouldn't this coarsing operation be usefull in Lw too, resulting in lighter meshes?
Zbrush SDS lvl1 meshes, Nurbs bakes, etc, would all benefit from this adjustment: from sticking to the surface, to, sticking to the surface when Subdivided no?

Cheers

EDIT: In the case of LW i think it could be also interesting to hava some way, if there isn't already, for transformations and operations over the mesh be calculated to the SubD surface itself (or at least the deformed cage) and not the SubD cage, which is not the surface final state.

Examples:
-UV Relaxing: UV correspondence is better
-Mesh Smoothing and Relaxing Operations. Spacing between polygons is more accurate.
-Snapping: You don't want to stick soemthing to the cage but the surface itself.
Etc...

jay3d
08-08-2010, 07:15 AM
A paper from 2006 is doing just that with the maths, now somebody needs to implement that in LW ;)

http://wscg.zcu.cz/wscg2006/Papers_2006/Full/B89-full.pdf

probiner
12-17-2010, 08:54 PM
Interesting, but that removes the edges, which is also a nice option. But..

My confusing post was targeting something like: add deformer to a mesh so that when it's subdivided the surface will pass by the original position of the vertices and not being shrunk, displaced.

This is something that will happen to a retopology mesh, or anything that snaps to a references mesh and after that subdivided. So you had all that work snapping something, when in the end it will be displaced.

The solution by now is to to dense meshes. But not a good one.

probiner
05-29-2011, 07:42 AM
Jay, i now see how could i use such a solution if i divide the Retopo mesh first.

Cheers

probiner
12-04-2011, 03:54 AM
this does it ^^

http://www.youtube.com/watch?v=tIeJsy-Tj3I&feature=player_embedded#!

JeffrySG
12-07-2011, 10:12 AM
this does it ^^

http://www.youtube.com/watch?v=tIeJsy-Tj3I&feature=player_embedded#!

I wish that guy made those plugins for LW and modo as well!

Sensei
12-07-2011, 10:56 AM
It can be trick relying on fact that sub-patch is generating always the same number of polygons. f.e. level 3 in cc is generating 64 polygons. So routine is searching for array of 8x8 polys, and they're/must be connected by edges. And so on with the rest.
Then tool knows where are borders of each sub-patch. And generates in loop experimental sub-patches, until the one, perfectly matching is found. Very time consuming task.
If that's the way it works, it won't work with mesh that has been edited after freezing. Especially if editing was not just modification of vertex position, but adding, removing, cutting geometry. In such case order of polygons is destroyed.
LW sub-patch (old one) algorithm is not available for people.

JeffrySG
12-07-2011, 11:10 AM
It can be trick relying on fact that sub-patch is generating always the same number of polygons. f.e. level 3 in cc is generating 64 polygons. So routine is searching for array of 8x8 polys, and they're/must be connected by edges. And so on with the rest.
Then tool knows where are borders of each sub-patch. And generates in loop experimental sub-patches, until the one, perfectly matching is found. Very time consuming task.
If that's the way it works, it won't work with mesh that has been edited after freezing. Especially if editing was not just modification of vertex position, but adding, removing, cutting geometry. In such case order of polygons is destroyed.
LW sub-patch (old one) algorithm is not available for people.

That makes total sense. I'm still always impressed by the knowledge and skills of all the plugin creators out there!

Chris S. (Fez)
12-07-2011, 11:11 AM
I can personally vouch for that Max plugin. Works quite well.

probiner
09-16-2012, 03:57 PM
I can personally vouch for that Max plugin. Works quite well.

Apparently he did it again :)
Not a script but an actual Max Plugin: http://www.youtube.com/watch?v=DXEg422xdWE

It also does, partially, ahaha: http://www.youtube.com/watch?v=ABnfOl_t4nE


He's whole youtube channel, makes you go... .

Cheers

Kryslin
09-18-2012, 11:36 AM
It appears that the max plugin in the first video looks for 2 x 2 groups, and reduces those. That's why it reduces by levels. That would simplify things a great deal. With a little bit of math, you can figure out how much those points will move. The trick is the direction (average, subdivided normal, maybe?)... (I've looked at this, on and off, trying to unsubdivide poser meshes. I had some limited success with it, but nothing great...)