Experiments in creating necklaces etc for use in Poser's cloth room
Back to the necklaces. Or more accurately, back to the chains. I'm intrigued by the difference in dynamic behaviour between (1) a string of rectangles connected by their short edges,and (2) astring of diamonds connected by the vertices with the acute internal angles.
Seems a goodidea to do some very basic tests.
To make sure that I use exactly the same dynamics settings and groupings for different chains I put four separate chains in a single prop. The chain-link mesh for each is identical. The vertices of the diamonds are where the centre of the edges of the rectangleswere. I made duplicates of the rectangle-dynamics and diamond-dynamics chains and rotated them 90 degrees so that the dynamic part was in a vertical plane. I assigned the first rectangle/diamond of each chain (green) to the choreographed group - this effectively fixes it in position.
Here's a screenshot to clarify the setup:
My first test was done in Poser 6 (yes six) - 500 frames,just letting the chains drop and swing fromthe green choreographed faces. It took about 10 minutes to run the sim, so around 1.2s per frame. I'm just rendering an AVI...
The video (all 17 seconds of it, like The Cure song- I shouldhave put that as the soundtrack, it goes nicely!) of the 500 frame Poser 6 sim is now up at YouTube here (no voiceover, commentary, or explanation - I hate those unless they're done well.And very, very few people can do that! The code is HIz1N1RiU3k if you want to type that into the normal YouTube URL, or right-click the embedded video,Copy URL,and paste into the browser's address bar. I hate embedded video!)
- And here are a few screenshots from Poser 6:
A brief aside regarding steel balls and springs...
@3dcheapskate - side view of that simulation here ( sQeZjeLXxgQ )
I found this view to be rather more enlightening.
Assigning the chain links (yellow) to the Soft Decorated group (instead of the Rigid) seems to get rid of the erratic behaviour from the diamond-based mesh. Also everything seems much slower - the furthest chain only does threeback-and-forth swings whereas it did about doublethat using the rigid group. ( 60Q8CSj694c )
Then tried assigning the chain links (yellow) to the choreographed group (along with the green),expecting them to remain fixed leaving just the red mesh to drop and swing. But nothing moved at all - the red mesh was somehow fixed in place too!?
SoI tried assigning the yellow to the dynamic group. Since each link is a mesh unconnected to any otherI expected the links to drop independently (there are no collisions set up),leaving the red mesh to drop and swing. That's sort of what happened... ( jiXKWKQd0sY )
Haven't done any more on this idea for a while, but yesterday I was looking at some Bajoran earrings on ShareCG and thought that was an ideal item to try this with. So I knocked up a quick, simple model of a Bajoran-ish earring - two attachment points on the ear, a dangly chain hanging between them, and another dangly chain hanging from each.
Result first (I forgot to adjust the render size for the movie format so you don't see her ear till the end, and then just the earlobe):
And three renders (showing the ear) of the first, a halfway one, and last frames:
And here's a Blender screenshot of part of the mesh:
The yellow is the chain links (assigned to the soft-decorated group), red is the simple dynamic mesh (assigned to the dynamic group), and the green is the attachment (assigned to the constrained group).
Re-render of the same sim with the camera moved a bit so you see the attachment points:
Running a different sim with her head tilting backwards throws up a problem that I didn't spot in the first sim - parts of the chain seem to stick together. This is probably because separate bits of my original mesh overlap (or are simply too close to each other) and the cloth sim treats them as if they were connected. I'm going to play with the mesh a bit to see if that's the case.
P.S. In PP2014 on my several years old Dell Inspiron laptop the first simulation (90 frames total, 50 frames shaking her head as it bends forwards, 40 more frames for the chains to settle a bit) took 95 seconds to complete, a tad over 1 second per frame. There's a bit of stretching and twisting of the chain links (they're in the soft decorated group) but it's acceptable for this test.
estherau last edited by
Looks really good to me. Congrats. very clever.
Tried to 'unstick' the adjacent chains by a slight adjustment that ensured there was no overlap between the U-shaped dangle and either of the two dangles with a loose end (except at/near the attachment-to-the-ear points of course). The chains still stuck together
If a slight adjustment didn't work maybe a big one will. I rotated the two free dangles by 90 degrees so that they were sticking out horizontally. That seems to do the trick!
It's not obvious from this video, but the top link in the front/lower free dangle sticks out unrealistically for the whole sim, whereas the top link for the back one doesn't. May be worth a look.
I was a bit off in the previous comment - it doesn't stick out unrealistically for the whole sim. It behaves quite nicely early on, but sort of gets stuck halfway sticking out at 45 degrees.
Anyway, I decided to try and add a chain connecting the two free dangles. I didn't expect it to work so I was pleasantly surprised...
(the mp4 is 320x240 because my laptop is getting rather hot rendering videos from Poser)
@estherau It's definitely showing a lot of promise. I think that loops like bracelets and necklaces seem to work best with a dynamic mesh of rectangles connected along their short edges, whereas freely dangling chains seem to work best with a dynamic mesh of diamonds connected by single vertices (provided you don't run into the chaotic behaviour). Dangly chains connected at both ends I'm not sure. The whole earing in the above videos uses a dynamic mesh of diamonds connected by single vertices and seems stable. But the next one (which I'm still working on) is rather chaotic.
Assigning the modelled chain links to the soft-decorated group results in some stretching and distortion, but generally looks acceptable, whereas assigning them to the rigid-decorated group is a total failure as various links appear to disconnect and dance about independently of the dynamic mesh.
PP2 files for three of these test earrings uploaded to ShareCG if anybody wants to play. BajTest4 is the original sticky one, BajTest5B has the free dangles starting sticking out horizontally so they don't stick, and BajTest7 has the extra chain connecting the two free dangles plus a couple of red gems replacing some links.
ghostship last edited by
@3dcheapskate This is some crazy s**t. You kn ow when your done someone's gonna ask you to make some nipple chains for them. LOL
@3dcheapskate Awesome work. Thanks for sharing your tips. Very much appreciated. Have to make a few tests too. Have some crazy stuff in mind too and those tips are for sure useful :)
@ghostship - hopefully when I'm done it'll be so easy for anybody to knock up a dynamic chain that they won't need to ask ! Talking of crazy s**t - I had a really bizarre sim result on a simple chain necklace loop. Remodelled everything from scratch and tried again - same bizarre result (and it only happens using the soft decorated - the rigid decorated works as expected). Rendering a video in a short while.
@Ladonna - From what I've seen so far the two things that cause the biggest difference in behaviour are (1) rectangular quads (single edge connections) v. diamond quads (single vertex connections) for the dynamic mesh, and (2) using the rigid decorated group v. using the soft decorated group.
Anyway, it's still ongoing, but meanwhile here's an aside - another possible use for this idea (and nothing to do with nipple chains! The proximity of this test to that suggestion is mere coincidence...):
Result of a 90 second sim (with 10 drape frames) positioning the mesh correctly for her arm on frame 1, and correctly for the wall plate on frame 90. Some obvious distortion of links at both ends, but this was just a quickly knocked up 'I wonder if that would work?' A little bit more thought and care about how to connect the constrained and choreographed parts to the chain/dynamic part should improve that.
Okay, that weird sim result. Here's what it looks like after 90 frames with the chain links assigned to the rigid decorated group - this is as expected:
But if I put the chain links in the soft decorated group - weird s**t! The video clips only 1 second (30 frames) but it continues rolling like that forever without moving.
New video uploaded, but this time about a minute and a half. A few tests of the latest earring, using both diamond and rectangular meshes for the underlying dynamics. This time I did a bit of video editing and put several 3 second animations together with some explanatory words and pictures.
Increasing the cloth density seemed to fix both the diamond-based mesh's overenthusiastic bounciness and the rectangle-based mesh's refusal to move.
Not sure how generic that observation will be. Scale may have a big impact (i.e. earring as opposed to anchor chain). The fact that both ends of the chain are fixed may also have a big impact.