Cloth Jumping/Flipping around
hey all, im having some cloth trouble. well, this whole project has been cloth trouble. i find it impossible to get the c4d cloth engine to give good results without an absolute battle. granted, im new to it, but still, there are only a handful of parameters to tweak. anyway... iv attached a scene file...
the problem is as follows...
-there are 50 frames of pre-calc to get the strap into the correct position.
-rigid sim on camera model starts at 0, camera falls, strap is attached via belt tags, strap gets pulled along
-as soon as the cloth sim starts, and the strap is pulled into position by the belt cube obj, you can see the cloth moving very erratically, close to the belt.
- around frame 0 when the camera starts falling, the cloth strap starts going nuts on BOTH sides, at both belt points. every frame it flips here and there to another solution, not giving a smooth predictable fall, as if being pulled down by the rigid body.
iv tried everything i can think of. so many settings tweaks i've lost count. jacked all the simulation settings to (what i think are) realy high values based on what the defaults are. iv tried making changes to the geometry as well... that doesnt help.
so at this point, im out of ideas. id like to know why the cloth is freaking out so much. too bad there isnt (as far as i know of) a general dampening setting, like the linear and angular velocity options in the rigid solver.
subdivided the long polys (the thicker bit between the end and the fat part of the strap), added a vertex map so some areas are stiffer (the transition areas particularly where there are vertex's so close to each other).
Turned down Iterations (more than needed). Increased Air Resistance, set Global Drag to a constant 1%. Turned on Self Collision in the Expert Tab and increased it's sub-sampling a bit, rotated the moving cube so it ends up the right rotation when it ends (don't know if that helps). Lowered Stiffness and increased Flexion (actually a form of stiffness) and added some Rubber so it can stretch a bit a take up some of the energy (all modified by the Vertex map). Decreased Bounce in the strap, you don't need extra energy in the sim maybe even 1% is too high.
Don't know if that's what you need and maybe Softbodies would be something to investigate but FWIW...
thanks for digging into it brian! i see you put BACK some things i had in previous versions, like self collision, higher sub samples, etc. Like i said i bounced around so much trying different things, the version i posted is just where i stopped, posted, then kept trying hoping someone would have better luck than me.
regarding the flexion, that one keeps confusing me. i click on things and bring up help when i dont understand them, but the wording on flexion was never quite clear. so i usually bring it to 1, or 100, compare the results to try and figure out what is making it stiffer or floppier, than i forget an hour later, and start again. my understanding was higher flexion makes it LESS stiff, as it is more able to flex. Lower flexion increases rigidity.... right?
interesting point on the rubber %... to absorb energy. Ill keep that in mind as i tweak.
as far as softbodies or other methods, i think my first idea was to do it with a dynamic joint chain, but then i know even less about setting up joints. The low poly cloth cage was something i used to use alot in Lightwave, so i knew it was at least a method i could work with and know basically what i was going to get, as long as the c4d cloth behaved. If i know a enough about cloth to get this far, i know even less about softbodies. but since this is most of the way there, ill keep going this route.
Ill fiddle with this for a bit and see what i come up with. Thanks for your help!
cloth was created for cloth on a character, collision with the character is important but self collision is not even turned on by default, and your sim requires it. Same for stiffness it was setup for clothing not with the kind of materials in a camera strap in mind (I'm guessing).
Your sim is more similar to cloth dropping in a pile, self collision is absolutely necessary but in reality cloth is billions (trillions?) of points colliding, it would likely be more accurate with tons of vertexes but the performance hit is massive...
Rather than bones I'd try to model it so it's already in position rather than moving it into position. Right now, if you look at the movement of the cube (up close) it jumps from frame to frame (like it has to) to it's next position (for that frame) and the cloth *jumps* to try and catch up, it's no wonder it has too much energy...
Flexion is stiffness, torsional stiffness - the image in the help for Flexion makes it clear (I think) it does get stiffer as the Flexion value goes up (not down)
thanks brian. good point on the cubes pulling it along. maybe if i slowed that down it would drop the overall energy. the reason i did it that way (flat to bend into position) was because it was easier to UV a flat object, and then let it get bent, so id be reasonably sure the UVs would be preserved and suffer minimal stretching. even if i modelled it flat and tried to bend it into position, i would probably screw up the UVs.
i did end up getting a decent solution after all. its not perfect, but its way better than the file i posted originally. i used a combination of your suggestions/settings, and a later file id been working on after my original post.
the one thing i kept fighting was collisions simply not registering. the strap would always end up intersecting itself. i tried turning up the sub samples, increasing the collision distance, and nothing mattered. i also, for a time, had 1 collision object simply stop calculating randomly. i would have to re-add a cloth collider to it every other time i pressed play it seemed.
so if i had to do this again, would you recommend soft bodies over cloth? or perhaps another method all together?
Collisions are a problem, yes. increasing the subsampling has to help (in the expert tab) but also increasing the number of points. It slows everything down but it's a better chance for accuracy.
I don't know about soft bodies haven't tried anything quite like that in a soft body. They seem intended for round things (or at least 'volumes'). However that being said they also handle collisions pretty well. I'll try a test just to see what happens.