Thursday, November 29, 2012

week 3 cont'd

week 3
it looks like i have the guide hairs locked in pretty good,
as it stands right now the guide hairs are being deformed by an attr transfer to the face from the
velocity&normals of hand instead of a sim. this gives me finite control of the
strength of the deformation of these guide hairs based on proximity and direction of the hand.
Then i gave my self some dampers built in to further control this

I am still having problems getting the actual fur to behave according to the guide hairs. below is an example of the fur only:

here we can see that the hairs pop from frame to frame both in position and in length, which is undesireable to put it mildly
from here i'm going to look into a few differnet methods for locking this down:
1)one thing i think that may cause this is that the area of the polygons changing from frame to frame based on
the deformation of the face/head animation.

2)my next potential trouble maker may be the connectivity mesh not grasping the attrs from the painted head mesh appropriately,
this might explain the length wildly changing in the manor it does

3)my last avenue to investigate would be the normals changing from frame to frame in an unpredictable manor
from frame to fram based on the overall rotation of the head/face geomerty. i have seen this before possible
solution in creating a vopsop that will measure the rotation transformation of the points on the face based
on quaternion interpretation

One thing that I think is kind of cool and is an added bonus, is that i created a psuedo volume displacement system  for the facial mesh based on attr transfer from the hand  to the face geometry. Here I use a color attr transfers from masking  purposes to define and active are of the facial mesh, a displace along normals based one a point cloud filter, in conjunction with the acceleration & velocity attributes of the points of the hand to determine acceleration and the origin of the velocity.I then use those maps to define a leading edge of the attribute transfer to create a bulge within that mask and the trailing edge of the attr transfer will stay depressed and then return to normal when the hand passees over the attr transfer region of the color mask from the hand.
From here the bulge and depresion on the repective leading and trailing edges of the attr transfer are multiplied by a painted attribute on the skin dictating which areas are over bone and which are more fleshy and will receive greater deformation values.
the end result, we get a cheat of a volume diplacement without simulation time that its strength is cheap on time, animate-able,  and adjustable on the fly, giving the artist the ability to produce more iterations in less time, for  just about anything but extreme close up shots.
below is a prelininary example of this volume displacement cheat at work:

the next step for this volume cheat will be to smooth out these deformations. Then replicate this in the displacement at shader level to give
smoother and more detailed reults to complement the initial sop level deformation.

No comments:

Post a Comment