collision deformer SOS pls..pls..pls help this child....

[Fleming Lin]

hello internet....I have been working on this collision deformer node for a long time and still cant get it right....I have done many research and even ask people for the reference code but IT JUST WONT WORKS!!!!!!!!...

the big idea for my collision deformer is

first do a intersection test between the collider bounding box and deformer bounding box

    if intersected, check if the collider bounding box contain the deformer vertex( in world coordinate)...

      if contained, do allintersection test for the vertex to see if it is inside the the collider mesh or not , using   

               (if (hitPoints.length() % 2 == 1))

and then do the deformation and save the point to the finalpnts array( used later to set all positions )

for debugging purpose I asked the maya to print out the collision point index, and it works fine, but the deformation algorithm I have in that same if loop just wont works........I have no idea why and really ran out of idea..please help....

here is the test result

here is the c++ code I have.

https://github.com/fleminglin/sos/blob/master/flinCollide.cpp

sincerely

flin

[Michael Boon]

What is not working? Your code finds the correct points, and prints out the correct indices, but it doesn't move them correctly? So what is it doing wrong when it moves them? Is it not moving them at all? Moving them to the wrong place? I see you calculating finalPnt and then entering it into finalPnts, but I'm not sure what happens from there.

Have you tried printing out the old and new locations of each point (in world space and/or local space)?
Have you confirmed that values set in finalPts are actually used to set positions of vertices in the output mesh?

[Fleming Lin]

hi Michaek, thanks for taking time reading my code... and the problem here is my code doesnt move the corresponding point on the deformer surface(sorry I should be clearer at the beginning)  in other words, nothing happened at all when I collide them(and this is the part drive me crazy, it's like maya is dead....)... 

Have you tried printing out the old and new locations of each point (in world space and/or local space)? 

yes I did, and I had created a MPoint difference = collision point - deformer point, and ask maya to print out the "difference.x, y and z" but all I got is 0,0,0.

And I'm sure the point I passed in the intersecoter.getClosestPoint is in the world space, however I'm not sure about what space the returned closest point is in, (I'm assuming it is in world space now, but I think I have done some quick test by multiply the closest point by the collider local to world matrix before, but still, nothing happened when collide.) 

Have you confirmed that values set in finalPts are actually used to set positions of vertices in the output mesh?

I'm sorry but I'm not quick sure what you mean here, I assuming after I update the finalPnts and use it in the iterGeo.setAllpositions ( or outIter.setAllpositions( just another MItGeometry iterator I created for the outputgeom attribute, I heard from someone else that this might solve the problem, but I have tried them both, none of them works) ),     the deformer surface will be update automatically?  ( and I have also tried fnMesh::updataSurface, but nothing happend...) 

thanks

flin

[Michael Boon]

We're not making sense to each other here :)

Are these points all where you expect them to be?

- deformerWorldPnt

- collisionPoint

- finalPnt

It might be worth colliding one cube against another, just because it's much easier to do the math in your head and know when the results are what you expect them to be.

Actually, in the code you shared, it looks like you use originalPnts to set the output mesh, not finalPnts. It looks like you intended to copy originalPnts into finalPnts and then use that to set the output. Could it be just that simple mistake?

finalPnts.set(finalPnt, i);
...
outIter.setAllPositions(originalPnts); // Should this be finalPnts?

[Fleming Lin]

hi Michael, 

thanks for the sweet advise! unfortunately, it has nothing to do with the  outIter.setAllPositions(originalPnts), I changed it back to finalPnts, still wont works. however, I just used two cubes to test my deformer, and get some interesting result.

here is the code

the result in perspective view

in front view, ( xy coordinate)

so it turned out I did something wrong with the intersector getClosestpoint...but even I stored the wrong finalPnt in the finalPnts array, how can the deformer mesh stays the same shape??(....maya...I dont know...maya.....)

sincerely

flin

[Michael Boon]

I have to admit I haven't written a modifier before and I don't see anything obvious wrong with what you've done.

It seems strange that you are using a MItGeometry outIter to set the positions. It would seem more direct to me to set the positions on your MFnMesh fnOutmesh, which controls hOutputGeom. I'm not sure if that's the problem but that's my best guess right now. Hopefully there are others here with more experience.

[Fleming Lin]

hi Michael! 

I just used the default MItGeometry that was provide in the deform method to set the final positions and everything works perfect now! My original intention was to used a different way set the final result, but I guess I was over complicated the code by doing this whole "MItGeometry outIter" thing. But still, thank you so much for your help, you are my hero of the day!!!!

sincerely

flin