PDA

View Full Version : Deformations with shape animation


Atyss
05-27-2002, 08:03 PM
Hello people,

Once again I'm struggling with deformations, and need some help.


Context
This time, I applied shape-animation to some parts of the hand. This shape animation is driven by a slider (wich control the bending of a finger).

I defined three shapes: one at rest, one half bent and one fully bent.

The problem
There is no problem... until the finger gets close to the fully bent state. The closer to this state, the more f@@@@d up the shape becomes, and gets back to normal right before reaching that state.

Here is a link to explain what I mean:
http://membres.lycos.fr/grimoireinterdit/img_demo/Enigmus_indexShapeInterpola.jpg

I tried normalize/non-normalize the Mixer, it doesn't change anything.

Am I missing something?


Merci - Thanks
Bernard Lebel

geoath
05-27-2002, 10:25 PM
hi Atyss!

i'm trying to figure out your workflow:
a) u enveloped the fingers with bones
b)created a slider for rotaning the bones
c)adjusted the fingers with the slider in each position (maybe also created clusters before) and created shapes for each pose
d) pluged the shapes weights (after u created clips in the mixer)in a way that when slider is 0 then halfbend and full bend shapes weights r 0 in -.05 halfbend shape weight is 1 and fullbend is 0 ect...(don't know if u used expressions or link deform).

well if let's say that this is the way u worked then i guess there should no be a problem..(unless i miss something too :) )

have u checked if for any reason (especially if u used link deform) the half bend shape weight "jumps" to "strange" value when it should be near to 0?..check that in the mixer by adjusting the slider and watching the half bend shape weight .
Maybe u should check that rest shape weight is near 0 too, cos even though in simple shape animation it doesn't affect the shape when normalize is unchecked even when it's weight is 1,due to the additive nature of the unchecked normalize, here it might create a problem since the shapes r applied on top of the envelope,not totally sure about that, i have to check it.

i haven't helped u a lot just tried to give some logical explanation i'll try it myself as soon as i'll be in front of XSI
best regards
george

geoath
05-27-2002, 11:49 PM
@@Maybe u should check that rest shape weight is near 0 too@@

i checked that and there is no reason to animate the rest shape weight..i was wrong: it doesn't influence the animation..

i made a simple example with a cylinder, following the workflow i wrote before and didn't face any weird results. (i used "link with" to control the shape weights through the slider).

if u used expressions instead of "link with" make sure that the weights of the two shapes mix smoothly from the halfbend pose to fullbendpose (u know what i mean: from -0.50 up to -1.00 value slider, fullbendshape reach it's maximum value 1, while halfbend shape gradually falls to 0 )

i still can't find what might created the problem to u..(it's always difficult when u don't have the scene yourself..creating a simple example isn't a solution)
Anyway in case u solve it please post the sollution..
Best regards
george

Atyss
05-28-2002, 12:32 AM
@@@@
a) u enveloped the fingers with bones
b)created a slider for rotaning the bones
c)adjusted the fingers with the slider in each position (maybe also created clusters before) and created shapes for each pose
d) pluged the shapes weights (after u created clips in the mixer)in a way that when slider is 0 then halfbend and full bend shapes weights r 0 in -.05 halfbend shape weight is 1 and fullbend is 0 ect...(don't know if u used expressions or link deform).
@@@@

This is exactly what I did!



@@@@
have u checked if for any reason (especially if u used link deform) the half bend shape weight "jumps" to "strange" value when it should be near to 0?..check that in the mixer by adjusting the slider and watching the half bend shape weight .
@@@@

Here is a small mpeg to show you what happens.
http://www.angelfire.com/film/grimoireinterdit/Capture.mpg

To explain a little further, my slider has a range of -1 to 1, 0 being the default state (corresponding to the state when the finger is in rest position). At -1, the finger is fully bent towards the inside, and at 1, it points up.

What you see on the movie is when the slider goes from 0 to -1.

As you can see, the problem of "deformation weirdness" occurs when the slider is very close to -1. But it works fine when it is above 0.


@@@@
Maybe u should check that rest shape weight is near 0 too, cos even though in simple shape animation it doesn't affect the shape when normalize is unchecked even when it's weight is 1,due to the additive nature of the unchecked normalize, here it might create a problem since the shapes r applied on top of the envelope
@@@@

I'm not sure that I understand what you mean. But from what I know it is always safer to put shape animation on top of the enveloppe operator.


Thanks for the reply!
Bernard

Atyss
05-30-2002, 01:44 PM
(thanks to Adam Sale)

The problem was that I had vertices that were "entering" inside the hand when the finger was bent. To solve it, I had to change the enveloppe and make those vertices to not cross other surfaces. In fact, it is better to make a flattened tube shape then deform it. It worked.

Bernard Lebel

HapZungLam
05-30-2002, 08:46 PM
I am not trying to hurt you. but it is because your finger (those joins) are modeled badly.

Sorry, i shounldn't use the word badly. should say incorrectly. Lets say, you are modeling a leg. the join on the knee should be modeled like, the front of the knee, all the edges are placing seperate. In the back of the knee, the edges should be close tight. I don't have a picture to show you here. I hope you can understand.

Like this *\ \ | / / *. You see what i am saying? the top would be the front of the knee. Botton is the back.

Same with the fingers edges. Doing this will blend nicely.

i hope this can help.

Atyss
05-30-2002, 08:57 PM
Well HapZumLang, if you look at the image below, I think that it is modeled as you said it should....

http://membres.lycos.fr/grimoireinterdit/img_demo/Enimgus_badFingerWire.jpg


Please elaborate.


Salutations - Cheers
Bernard Lebel


(BTW: The other fingers joints are not "correct", I know, but they do not pose deformation problems.)

CGTalk Moderation
01-13-2006, 08:00 AM
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.