I have been trying to work with a lattice to deform an eyeball for a while now, and it is really sucking big time.
Is it just me, or is it a little counter-intuitive to deform a circle using a rectangular lattice?
Maybe it is because I’m new to this, I don’t know. But anyways - here is what I was thinking would be WAY easier, and in my opinion more powerful / effecient / etc.
Say you have an object of any shape. The most intuitive way to deform it, in my opinion, would be to have a deformation box / lattice of the same shape. So wouldn’t it be handy if you could duplicate the object you wish to deform, scale that duplicate to a slightly larger size than the original object and then press some button “convert to lattice”. Now you would have a lattice that has vertices in logical positions allowing for easy intuitive control of the deformation.
What do you think? Should I post this as a request on Blender.org?
Or, even better, can you already do this somehow? If so, please explain!
I was thinking it would be good if you could deform the lattice, THEN parent the object to it. At this point the object should not be deformed. Then any further deformation of the lattice would deform the object.
Or this already possible, and I just don’t know about?!?
I just tried. If you deform the lattice and THEN parent it to the object, the object instantly takes on the deformation performed on the lattice before you parented it.
Besides, I don’t think that that would be a LESS complicated solution. Personally I would rather press SHIFT-D, S, and then “Convert To Lattice” - rather than try to manually deform a rectangular lattice into some other shape.
what about this: you use the lattice normally and deform it to the shape you want before parenting the object to the lattice. Then the deformation that is applied to the object is any further deformation of the lattice, rather than the deformation from it’s original shape.
I’m not sure whether the maths is still very complicated though, or whether it would be that useful, any thoughts?
I use deformed lattices all the time. I use them for target shapes on NURB keyframing, I use them for animating hair , cloth and drippy stuff. I use them for wierdly shaped eyeballs and for bouncing fleshy body parts.
I use them when I have thermal material deformation to illustrate, or lately to do vortexing on a set of wings I am goofing around with.
To say:"If it is not a square grid the math get incredibly complex and the result is well, probably unusable. " is not something I would agree with.
For what its worth here is an example of that sort of thing in action in one of my old files.
Nice Blend file, and model but the Lattice is nevertheless a rectangular-based lattice, heavily deformed. What nuance was saying is
to have non-rectangular lattices from the beginning, which would be difficult.
What I meant is that the lattice undeformed structure must
be rectangular for an easy non-singular mapping of the space. Probably
we can think of cylindrical or spherical lattices since these too are coordiante systems but not much more than this.
If nuance wants few control points to control a shape
and wants these control points in the ‘shape of the shape’ then he
should use SubSurf, imho.
I just left a post this morning in the hotkey thread (in news and chat) that shows an object deformed twice over with lattices. Here’s a better example:
arangel: Thanks (Blush), The model is from clay. Its a rough model I made to guide me in visualizing the lattice construction. (Clay/Dsculptor generated mesh, imported to Blender as .obj)
The file is called “shrinkwrap lattice to express the way I made it.” Used a lattice with what looked like a good number of divisions and first selected groups and scaled them together until I had a tube and then tweeked the tube around the referance torso the way you might a mesh. Took four hours.
The anim is just dragging a mesh through the “influence zone of the lattice.” Select the mesh and move it around the area a little bit and you will see what I mean.
S68: Hope you did not take that as busting your chops, I see your point better now. My Mentor is after me to work out lattice structures as fullerene structures. but I am as yet unconvinced that they would be worth the trouble to program since you are doing the same thing with grouped cubical structers like in this example anyhow. The argument for it is exactly the words you used. If you are mapping space “non-singularly” you are going to see opertunities in your accidents that you NEVER would have been able to visualize, and you could compound harmonics which might take modeling to a whole new level.
Easier said than done.
but the Lattice is nevertheless a rectangular-based lattice, heavily deformed. What nuance was saying is
