Skin Wrap Deformer Prototype Here, Help Needed!

[shikung444]

Hi guys, I'm working on a skin wrap type deformer here in sops and vops. I have attached a hip file with explanation of the problem, and what I need to happen.

I think I need to calculate the normal, and use some kind of rotation matrix to rotate the geometry using the same equation one would use for a bend. I haven't quite grasped what is needed to do this yet though.

If anyone can offer any suggestion I would really appreciate it! Thanks!

deformation_problem.hip

[symek]

Maybe Simon's work can help you:

http://www.sidefx.com/exchange/info.php?fi...p;versionid=146

cheers,

sy.

PS

Look also on the code here:

http://www.forum3d.pl/f3dbb/index.php?md=t...rst_unread_post

Edited March 11, 2008 by SYmek

[shikung444]

Hi Symek,

Thanks very much for your reply. I will take a look again at those resources. I can understand and make things in vops , but still haven't quite gotten my head around VEX coding (or C++). I will be digging through Simon's skin wrap vex code and trying to reimplement it in vops.

I also will be watching CMIVFX's VOP training DVD again, as I think I need to come to grips with the concepts there to understand the geometric transforms involved.

[sibarrick]

That one may have been made in vops, if the vop network isn't included let me know, I'm may have it somewhere. I can't remember if it used point clouds or not, if it did it might not have been vops, if it didn't it should do, since that would be the best way to do it

Are you doing this to learn vops or do you just need a wrap deformer that works well. If its the later you might want to check out this one too Wrap deformer its done in the hdk so its a lot quicker. Although the vex one isn't bad concidering what it's doing

[shikung444]

Hi Sibarrick,

I saw that some of the vex code was converted from a vop network, if you have the original Vop network that would be fantastic! There were a few vex operators there... some of them did use pointclouds, pciterate, etc. I will try the HDK wrap deformer as well, but I would like something that I have total control and understanding of, as I have a complicated and.. different setup planned and might need to customize and/or troubleshoot it.

I have heard that Vex and vops can process as fast or faster than C++ in the HDK... but here you said the HDK version is faster. Is that because of the way it is written, or because of some limitations of Vex or vops processing here?

Thanks so much!

P.S. I use the 64 bit windows version, I'm guessing that is compiled for 32 bit?

That one may have been made in vops, if the vop network isn't included let me know, I'm may have it somewhere. I can't remember if it used point clouds or not, if it did it might not have been vops, if it didn't it should do, since that would be the best way to do it

Are you doing this to learn vops or do you just need a wrap deformer that works well. If its the later you might want to check out this one too Wrap deformer its done in the hdk so its a lot quicker. Although the vex one isn't bad concidering what it's doing

Edited March 11, 2008 by shikung444

[sibarrick]

Actually it looks like that one was entirely vex code, and yes it does use point clouds. I think at the time I couldn't get point clouds working in vops. This sop is way easier in vex code, in vops it would get very messy.

In the hdk I was able to make a capture and deformer as two seperate steps that isn't really possible in vex as you can't add the necessary attributes. That's what makes it quicker. However now days you might be able to do the capture bit in python in which case it could be all as quick, not tried it.

Yes the compiled versions of the hdk stuff is 32-bit only at present until I get a 64-bit compiler....

[shikung444]

Vops may be messy, but I can understand them better than code at the moment

If I could calculate the rotation from the capturing point needed, I could try to continue developing another one as well. I might be able to calculate normals on the low res geometry, then use that info to modify the position attribute further.

I think the way I did it is already too slow though, I used a vex vop to turn the transform position into an attribute, then used attribute transfer to copy the attribute to the high res geometry, then used a point sop to add in the deformation. I can do the transformation in vops, but I am thinking to try to do the attribute transfer in vops instead as well. I haven't gotten my head around using point clouds yet (other than using the scatter sop)

[sibarrick]

I would recommend reading up on point clouds, they really aren't that difficult to use and are amazingly powerful for all sorts of things.

[shikung444]

I have finally finished digging through the skin wrap HDA, recreating it piece by piece. It's very impressive! Thanks very much sibarrack. I will also be experimenting with point clouds in the future.