Hi yall,
I have started a new project that I need some tech help with. First and foremost I am a glass sculptor, (glass blower to most). For the last 6 months or so I have been using blender to model new glass works. It has been a huge help in showing my customers the outcome before I do all the work. I have been asked to make tutorials of creating glass scluptures and here in lies the problem.
I need a slow moving fluid, something like cg honey. http://www.marshglass.us/tutorials/foldtut.html Here is a link to a video I did that will give you an idea of what Im wanting to accomplish with the cg. Sorry the video is of low quality but if you watch it all the way through you will see what I want.
One of the problems with an actual video for the project is the lighting. there is a bright light that cameras dont like. And Im limited to only one view.
Iwant to be able to view it from differant angles and such.
I have tired using meta balls with some success but I would really like it if some one with more experiance than me could steer me in the right direction.
Thanks Blender world
MIke
I don’t know if I have allot of experience, but what about this:
Make a fluid simulation with high viscosity. Add a forcefield near the source. (so the liquid comes out fast but slows down as the “glass” hardens) Remove gravity. Add a color ramp for the particle system so it changes color as the “glass” cools.
Thanks Mack,
That is exactly what I want to do. I have thought of that too. I just dont know enough about fluid sims yet. I guess I need to study more. Know of a good flluid tut?
I think you could do this with shape keys and object substitution if you don’t mind working with rather high-density meshes. I don’t think the fluid sim offers enough control over the “substance” aspects of a fluid object to have the material go from semi-fluid to solid as molten glass does when it cools. It’s both fluid and solid, which I don’t think you can do with the sim.
Using shape keys you can sculpt the progress of the form being built. Because the folding process has times where the molten glass adheres to itself & forms a new solid surface, you’ll have to do a transition from shape key sequence to shape key sequence, using different objects at each stage. The shapes can portray drawing out the glass and folding it in on itself, then when the glass would flow into a new solid form, a new object sculpted to match the current state of the first can be substituted in, with shape keys for the next folding step. This way, mesh intersections which would mar the "reality"of the look could be avoided.
A basic armature would also help to construct the necessary shape keys, since the folding is not a trivial motion.
Blending between the various object states could be done using layer substitution but that risks a “jump cut” if the match between objects at the “crossover” point is not perfect. Instead, the sequences could be rendered separately and a short cross-dissolve between them made in the Video Sequence Editor or Compositor, for a smoother transition.
I mentioned high density meshes because of the stretching involved. To maintain a smooth glass-like surface during the stretching part of the process, you’ll need a mesh that begins with high density, because the stretch spreads out the loops and edge rings, which could lead to less than smooth curves at full stretch, even with subsurface enabled. But the optimum mesh density would have to be determined via testing, I think.
Getting the feed rod right also presents some issues, but could be handled in similar fashion, I think. As is the case with many animations, the relationship between feed rod and formed shape would be reversed, with the shape controlling the feed rod position rather than the other way 'round.
Thanks chipmasque,
I like that idea. I had thought of the shape keys myself but not to the extent that you have. And the object substitution is a good idea too. That would have to happen about every 10 frames or so, dont you think? Maybe more in high detail areas. Elaborate a little on the armature idea. Im not sure I see how to use it.
Thanks
Mike
Folding a mesh back on itself using sculpting might be a lot of work, but an armature could do it fairly easily. The bones can be set to stretch (infinitely if need be), or you can scale long the Y axis for that as well. I think I’d start by using the armature to animate the draw-out and folding, ignoring at first the finer shaping. Then you could add shape keys that act to “correct” the shape to its final form just as is done with problem deformations at joints in a character animation. Once the object has been folded back on itself (you can allow the mesh to intersect some at this point), you can duplicate the object, apply the armature modifier and the final shape key on the dupe, then edit the new mesh to remove the intersections, resulting in a fresh shape to begin the next step in the process. If you mess up the new shape some in removing the intersections, you could use a Shrinkwrap modifier to clean up the parts that don’t match the first shape as well as you want – you can use a vertex group to control how the Shrinkwrap works on the mesh, so it only applies where you want it to.
I think an armature with only three or maybe four bones should work OK to get the stretching and folding done, so it wouldn’t be difficult to make one for every object in the transition series. My thought is that a new shape for every fold made in the process should suffice, since each fold is where the shape consolidates itself into a new surface.
As a test of the process animations, you could do OpenGL renderings of the transitions to check on various things like timing, match between transitions, etc., and when it all looks good, render it up with you material and lighting effects. That would save tons of time in getting the animation pinned down. I don’t think animation playback in the UI could proceed at decent speed because of the mesh density you’ll probably need, though. Maybe good for a rough check, but otherwise too slow, even on a faster machine.
I realize this is all “manual labor” in terms of the animation style used, but a sim would have to take into account the variable properties of molten glass along the temperature gradient, and I really don’t think Blenders up to that quite yet.
Funny thing is, back in 2.49b I was experimenting with the cloth sim & came across some settings that looked a great deal like taffy being pulled, and with vertex group weighting and pinning possible, might actually do better than the fluid sim. But it’d be a real experimental procedure, and I think that armature + shape keys would in the long run prove less work.
Another thought: If you plan on doing more that one folded shape for the demo, you can use various tools to alter the look of a base set of objects, so you can use the same animation and shape keys for each additional folded shape, but with some modulations so they don’t look identical. Additional shape keys that act in concert with the base set can do this, or even some additional animation on the armature, applied via the NLA Editor. Much depends on how complicated you want the demo to be.
Ya, he’s right about fluid simulator not working.
Instead of using an armature, you may want to use a curve.
You can make a shape key for the curve that acts like the motion of feeder and the molten glass. You can then add a cylinder like shape with a follow path constraint.
Evaluate the path and keyframe the cylinder’s offset to control the glass flow.
You can then add a second shapekey to the cylinder to change it’s thickness or have other affects.
Edit:
I had another idea. Instead of adding a shapekey to the mesh. You can add a lattice that has a curve deform modifier and follow path constraint. If the cylinder and lattice are in sync, then you can control the thickness and other characteristics that way.
@ macktruck6666 – good alternative ideas! The only drawback I can see to the second one, using the curve and mesh deform, is complexity – these things interact in strange ways sometimes. I used the same approach for a project involving a number of deformed cylinders/tubes, and the interaction of the two deformation controls turned out to be less than predictable, I got a lot of odd shearing off the curve path and unusual squashes in ways not expected from the mesh deformer. I guess maybe they weren’t “in sync,” but how to do that is not immediately apparent when setting up such systems. An armature & shape keys just seemed more reliable in terms of “tried & true” methods. If you have a lot of experience with the other methods, dealing with the unexpected might be easier, but if not, it could be more time-consuming than needed.
I really thank yall for thinking through this. It is helping more than you know. tomorrow I will start implementing the shape key ideas and see what I come up with.
Just a small mention of Booleans, since this seems like one of the pretty limited instances where they might be useful for animation.
The results of two Union-boolean’d objects with a Subsurf on top gives a result that looks a lot like metaballs, but can be controlled much more precisely (and combined with any other mesh deformation method).
This is with and without Subsurf, to be clear about the tris that result from a basic operation.
That’s a cool Boolean demo, but doesn’t it require two discrete objects? glassdog needs a similar effect to happen at the intersection of the mesh with itself when it’s fully folded back. This is what makes a series of shaped objects necessary, since the fused glass creates a new continuous surface. If the intersections could be masked (hard to do, I think, with a transparent material that will likely use RayTrans with a high IOR), it would simplify things, but if not, a new surface needs to be generated to accommodate the fusion. But I haven’t used Blender’s Booleans, so I can’t say if they’ll work with self intersections.