Relax SOP tool

[vi_rus]

Hi guys) Check out my new tool for smoothing geometry. Main problem of smoothSOP that after some iterations it almost don't smooth geometry. And speed of course.

[kare]

Hi guys) Check out my new tool for smoothing geometry. Main problem of smoothSOP that after some iterations it almost don't smooth geometry. And speed of course.

Its really cool dude..

Compared to smooth sop it is fast..

keep going dude:)

[vi_rus]

thanks, dude)) I promise, I'll keep going)

[JoshJ]

awesome, I was just about to post something asking about a better smoothing function. Any chance you can let us know the logic behind your network or code? Thanks.

[vi_rus]

Yes, of course) Logic is pretty simple... It just find neighbors for every point and compute average position. And user can set depth of neighbors. Neighbors in this context means points that connect with current point. That's all)

Edited January 7, 2012 by vi_rus

[vi_rus]

So... I upload my relax tool in exchange... You can download it here

http://www.sidefx.com/exchange/info.php?fileid=885&versionid=885

P.S. Added feature for distance blending when point attribute using.

Enjoy)

[kgoossens]

Hi guys) Check out my new tool for smoothing geometry. Main problem of smoothSOP that after some iterations it almost don't smooth geometry. And speed of course.

I see you have a very low cutoff Frequency in your smoothSop. Try cutoff Freq at 2.1 and Smoothing iterations 500

[vi_rus]

yeap.. you're right) now it works much better) don't know why I don't try it before... Thanks a lot for this advice!

Edited January 8, 2012 by vi_rus

[Pancho]

Hi there,

 

since I'm new to H coming from SI I struggle a bit to make this compatible with H14. In SI I made my own SmoothDeformer within ICE. So the general idea is for sure the same, but H14 spits out some error messages. Not sure how to get rid of them. If somebody could update the otl it would be great.

 

P.S.: How can I alter the otl since everything is grayed out? Is this some kind of compilation into a more powerful code? Are there any tuts which do explain how to go back and forth with this kind of network?

 

Cheers

Tom

[kgoossens]

If you right click the node you can set it to "allow editing of contents" then you can start to alter it

[Pancho]

Ah, thanks. Then I'll give it a try and see how far I'll come.

[Jason]

Just FYI, you can plug in the geometry into both inputs of an AttribTransfer and transfer P, which does about the same thing.

[Pancho]

O.k., can't get it to work so far. When I import the otl I seem to end up with three nodes. Neighbour Relax, Point Relax (Manipulate) and Point Relax (Digital assets) which can be found while pressing TAB. Point Relax (Mani.) then gets altered by myself, since there are at least two errors in the script because of the advances in H. In Neighbour_finder1 I edit the script from

 

float neigArr[MaxCount]

 

into

 

float neigArr[]

resize (neigArr, MaxCount)

 

and in the foreach > neighbour_relax1 I change all arrayLength into len.

 

With these modifications I don't get any more compiling errors.

 

BUT, so far only one point of the mesh moves (not necessarily into the corect smoothing direction) and when I open H again and load the scene I get an error: Couldn't remove old operator type definition sesi_temporary::Sop/relax.

 

With "Allow editing of contents" enabled I saved the operator type (not possible after "match current definition") after my changes.

 

To summarize everything:

 

a) no mesh smoothing so far

 

and

 

the error message.

 

So if anybody could take a look at the otl or tip me off in the right direction I'd be really glad.

 

Cheers

Tom

[anim]

as a quick fix you can chnge Neighbour Finder Code to this to work with H14:

sop
neighbour_finder(int MaxCount=0)
{
int neigCount=neighbourcount(0, ptnum);
int i=0;
float neigArr[];
resize(neigArr, MaxCount);
for (i=0;i<MaxCount;i++)
    {
    neigArr[i]=(float)neighbour(0, ptnum,i);
    }
addattribute("__neighbour",neigArr,MaxCount);
addattribute("__neighbourCount",neigCount);
}

[Pancho]

THanks! That seems to work! (0, ptnum) was still (ptnum, 0) in my version. Seems that this context also changed.