Make Aim Constraint respect deformed position? How (2)

[thumper]

I've got a fish character with lattice, bend, twist and motion path flow deformers and his eyes are controlled with Aim Constraints. However if I click on his eye while he's being deformed I find the eye's axis is no longer within the eyeball. Instead it's sitting in space where it would be if the deformers were turned off. Consequently the aim constraint is evaluating as if the eye was located where it's axis is, not where the deformed eyeball itself currently is.

So the eye's orientation is wrong. It's oriented as if it were sitting where it would be if it wasn't being deformed.

How can I get the aim constraint to evaluate as if it was sitting at the centre of the deformed eye, where in fact it should be?

Thanks, any help greatly appreciated.
John

[japunto]

I'm not certain about your setup, but often you can fix problems like this by altering the deformation order of your eye mesh. If, for example, you're rotating the eye via a joint that it is skinned to (which is aim constrained to something out front) try placing the skin cluster at the front of your deformation order. This will make it evaluate the rotation before the bend, twist, etc.
Like I said, I'm not sure if this is your real problem, so let me know more details if that doesn't help.
-JP

[thumper]

Thanks for the response JP.

Actually I'm not using a skeleton at all, just non-linear deformers. The fish is made of several nurbs surfaces grouped together and deformed by a lattice. Then that lattice is deformed by a sine deformer, several bend deformers and a twist deformer. Finally the group is animated along a path and deformed with Flow (local Effect). Each eye has an aim constraint. You can see the deformer order here http://www.rowes.com.au/EyeDeformerOrder.tif
Note the aim constraint appears above the other deformers and is separated by a line. Unlike the other deformers, it's not possible to drag it anywhere so I can't change its deformation order.

I guess its not really a defomer, but I wish I could get it to evaluate after the deformers instead of before.

Any ideas?

Thanks,
John

[japunto]

John, are you aiming the eyeRight transform directly? If so, you may want to break that connection, skin the eye to a joint and then aim that joint with the aim constraint. This will allow you to place the skinCluster before the others in the deformation stack.
Because the joint won't be deformed, you'll get an offset between the eye aim origin and the mesh, but everything should still work. Meaning, you'll be able to move your aim object around and see the eyes rotate around. If this isn't doing it still please take a screenshot of your mesh
in pre- and post-deformed states.
-JP

[thumper]

Hmmm, yes I am aiming the eyeRight transform directly. It does actually work this way but it gives me the sort of offset between the eye aim origin and the mesh that you're talking about. In fact that offset is what I'm really trying to avoid.

What I'm really trying to do is get Maya to evaluate the aim in the eye's post-deformed state so the deformed eye is looking at the target. At the moment that offset results in the eye not looking at the target but rather off to one side of it. The problem is this has the annoying by-product of making the fish look a little cross-eyed at times. I've tried moving the target further away but it doesn't seem to help. Perhaps the easiest way is just to set up a separate target for each eye and diverge one from the other when it becomes a problem.

I'm a little surprised though. I thought there'd be a way to get Maya to evaluate constraints after the deformations instead of before. I suppose though it's catch 22 - if the eye's orientation changed after the deformation you might end up with the eye's deformed shape not conforming to the eyesocket's deformed shape. I suppose Maya's current evaluation order is definitely the lesser of 2 evils.

[japunto]

Yes, the use of two aims which are children of an overall control is the way to go as far as avoiding crossed eyes. As far as the offset goes, I don't know of a way to avoid it. The problem is that you're aiming at something that isn't being deformed, so there will always be a difference.
Best,
JP