PDA

View Full Version : dynamically rigging this kind of hair?


amannin
04-23-2008, 02:44 AM
http://i28.photobucket.com/albums/c236/amannin/closeup_front.jpg

his hair is essentially giant leaves and all i want them to do is move as if the wind was blowing them (not a lot, just enough to give them some life). How should I go about doing this?

I'm trying nCloth at the moment, but since I'm new to it I'm not having much luck as of yet. I can get the ends to remain somewhat stationary (mind you i'm only working with one piece at the moment) but I can't get it to keep its form.

I've also tried a dynamic IK spline (influenced by a hair curve), which works great actually, however, it only responds interactively -- I would like for it to respond to fields...

Any help would be greatly appreciated.

PS. also, the most speedy of ways would be great, since nCloth likes to take its time :)

john_homer
04-23-2008, 03:59 AM
hi,
nCloth should be able to get stiff enough...
make sure you have the correct solverScale
set you nuclues subSteps to at least 30
apply on of the presets (like heavy denim) and crank the stretch and compression up to 200 or so... just keep cranking....

If you cant get it stiff enough with the properties...
I would not recommend using rigidity as it is buggy and would not work if you have combinned your shells (leaves)
instead, select all the verts in 1 leaf, create a component to component constraint with a max distance that is about half the lengh of the leaf, crack strength to taste, repeat for other leaves..

.j

amannin
04-24-2008, 07:27 PM
thanks for the input -- however since there are 18 pieces total, nCloth with collisions is terribly slow (i'll have to continue testing it another day).

for now, I just decided to use Maya's hair dynamics combine with dynamic IK splines -- the turbulence and other fields seem to do a decent enough job of moving the hair around with out too much penetration (it's not a very big shot, so i can get away with some)

thanks again for your help, but i'll have to try it another day when i have more time.

john_homer
04-25-2008, 12:15 AM
interesting....

what do you consider slow? (just out of interest)

I would expect that to solve in 10 seconds a frame.. or less... depending on how you are making it stiff...

.j

amannin
04-25-2008, 08:18 AM
well, with one piece it was near real time, with 18 however, I believe nearly a full minute went by before i had to hit escape. (however, that was with the tshirt preset, albeit, slightly stiffer) -- i had already started with the hair dynamics though by the time you replied, so i didn't try the things you mentioned...not yet at least...

SPECS:

E6600 (2.4 duo core)
nVidia 8800 GTS 640mb
2 gb Ram
WinXP 32 bit
Maya 8.5

I was also thinking about trying it without collisions, but then I thought to myself I might as well use Hair Dynamics, as it would be faster anyways.

I can upload the character if you are interested in giving it a shot, see if you can get the simulation down to a reasonable time...

thanks again

john_homer
04-28-2008, 12:37 PM
sounds a little odd to go that slow...

if you can post it I can try to have a quick look next time I'm waiting for a sim.

.j

toho
04-28-2008, 01:48 PM
May be combine 18 parts into one?

amannin
05-01-2008, 06:55 PM
http://rapidshare.com/files/111801114/robot_dynamic_test.zip.html

the outliner is a little messy, but the hair pieces are grouped together under hair...

thanks for the help.

john_homer
05-02-2008, 12:39 AM
Hi,
I'm not sure you uploaded the correct scene, there is no cloth there.

i'm not setting it up for youi, but I will give you these tips...

you seem to be working at a scale of 1:10 so set your solver scale to 0.1

if the meshes you sent are the ones you are trying to simulate, thats your problem.
make low res versions and wrap your hiRes to them,
currently they are double sided and around 2000 , I would single side them and make them about 20-30 and just turn the thickness up.

you also have lots of history on the meshes, delete ALL construction history before converting to cloth.

.j

wigal
05-02-2008, 11:25 AM
as toho suggested its always better to combine meshes that can be combined into one... it will speed up things alot...

amannin
05-02-2008, 05:47 PM
yah, i didn't add cloth to that version and apparently it still had all the history from before. but thanks for the tips, that makes a lot of sense and seems like it will definitely help. i'll try combining them into one object too to see how much that speeds it up.

two questions though since it will be a couple of days before i can go back and try this:

should i triangulate the lowRes meshes? and, since my meshes tend to intersect towards the roots, won't this affect collisions (with the other hair)?

john_homer
05-05-2008, 12:16 AM
should i triangulate the lowRes meshes? (with the other hair)?

HELL NO!
read the docs ;)

since my meshes tend to intersect towards the roots, won't this affect collisions (with the other hair)?
is best to not have them intersecting.. however, I suspect you can get away with this...

select all your base/intersecting verts AND the head (nRigid) create a point to surface constraint (ei, attach the hair to the head) now turn on "excludeCollisions" on the constraint.

.j

amannin
05-05-2008, 05:40 AM
i am currently experimenting with acheiving the right effect with nCloth, just in case i can't figure out the dynamic IK issue (refering to my other post)

so far trying to acheive the same effect, (mimicing what i did with dynamic IK) is proving much more difficult than i expected, but at least i'm seeing results...

also, it's interesting that in Maya, increasing the substeps increases the calculations (making it more accurate) and thus slowing it down, whereas with RealFlow, it is just the opposite. You increase the substeps which lowers the accuracy but speeds up the simulations.

mascio
05-05-2008, 05:28 PM
I've done a little scene.
May be it could be useful.

amannin
05-06-2008, 11:41 PM
mascio - seems like an interesting rig, however, the cloth always falls infinitely with gravity turned on. i tried point contraining the pieces to a dummy object but it only does 1 point which then shoots off into infinity when i press play - the wind attribute, with gravity off, seems to work fine though.

however, with wind, it ripples like it's made of cloth -- and if you've ever notice leaves on a tree, they tend to flap around while bending (they look pretty rigid, yet bendable), which i have yet to make nCoth mimic. On a previous test I tried cranking up the rigidity and applying a turbulence field, but it didn't work (the nCloth didn't respond).

unfortunately i'll have to invest more time into later; i need to finish my scene with my current setup (dynamic IK spline)

CGTalk Moderation
05-06-2008, 11:41 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.