Experiments in creating necklaces etc for use in Poser's cloth room


  • Poser Ambassadors

    @eclark1849 - you're probably right; I prefer dynamic cloth to conformers but my cloth room knowledge is poor ;)

    This looks like being an interesting thread ...



  • @eclark1849 "Shouldn't your dangly bits acutally be in the Rigid dynamic group? And Collision offset should be turned off or set at zero."

    As opposed to the 'Soft Decorated Group'? I can't actually recall why I put the chain mesh itself (the yellow bits) in the 'Soft' group, not the 'Rigid' group. I thought it was because the individual chain links became separated from each other as the underlying dynamic mesh (red bits) stretched, but that doesn't seem to be the case. I just tried the simple 128 link necklace that I uploaded (link in OP), but moved the actual chain mesh from the 'Soft' to the 'Rigid' group. The end result was very similar - but using the 'Soft' group it took about 90 frames whereas using the 'Rigid' group it only took about 20 frames !

    Minimum collision offset is 0.1 - it won't let me set it to zero. But collision offset/depth only affect the dynamic groups (not the constrained, choreographed, soft decorated, and hard decorated), don't they ?



  • I just wanted to remind people of these very, very, very, nice python scripts:

    http://www.morphography.uk.vu/~cagepage/looper/loopy.html

    There are several tools here that jewelry makers and others would surely want to use. The "chain" scripts make user-defined chains with good geometry that can then be clothified or can be generated "as is" for realistic chain draping and the like. I've used most of them, already, in P11 and they still work fine. (A couple hang, IIRC, but the others still give useful alternatives to them.) I haven't used some of those that have capabilities which I already have with other functions/programs.

    There are several chainmaker tools out there that are also free. Most do not have user-defined geometry, however. THESE DO! You can make the links as big or (almost) as little as you want. IIRC, though, you can't make a wide variety of chain "styles" like herringbone et al. But, with the parameters included, you can make any sort of "roundish" or "oval" chain link system you want. Some scripts include dynamics features, by the way, so those simulating draping would probably like those. (These scripts have been around a very long time.)

    In response to chains in the Cloth Room: I've tried it a few times. But, it seems like a lot of calcs for what it does if one is trying an actual physical simulation for chain links. Though, there are some that would really like to render draping jewelry/chains/etc, so to each their own.



  • @eclark1849 - "@caisson I'll need to check to see, but my think was that the dangly bits be rigid, but have zero dynamics, while the necklace would also be rigid but have some fold resistance"

    If on the other hand you're suggesting that the underlying dynamic mesh for the dangly bits (i.e. the white bits in the attached image below) should go in the rigid group, then I'd disagree. Each dangle is six chain links, and I actually want them to bend and twist, which is what they don't appear to be doing. Off to try some tests using different Dynamics Controls settings for the white bits...
    0_1486005801410_Untitled.jpg



  • @caisson said in Creating necklaces etc for use in Poser's cloth room:

    @eclark1849 - I think if the chain is all in the rigid group then it wouldn't bend at all i.e. the whole chain mesh would retain it's shape from frame 1?

    I did a really limited test ages ago with part of an object as soft/rigid, kind of like a collar. It was a completely separate unwelded mesh as per the manual, & in the soft group it did seem to both follow the underlying mesh & deform with it without detaching. Was a very quick test though.

    Each individual chain link in my necklace is a separate mesh, not connected to anything else. So if the yellow parts of the mesh (the actual links) are put in the rigid group, each link is dragged along with the corresponding rectangle of the dynamic (red) mesh. So the whole necklace deforms but each individual link maintains its shape.



  • Taking the original necklace-with-dangly-bits that I rendered in the OP, but putting the underlying dynamic bits for the dangles into a separate dynamic group as noted a couple of posts back wherethe Blender screenshot shows them in white.
    The chain link mesh (yellow) is all in the 'Soft Decorated Group', and the underlying mesh for the main necklace (red) is in the 'default' dynamic group. The new dynamic group uses default settings.
    The numbers in black are frame numbers:
    0_1486009922766_Dyn2grp.jpg

    However, if I repeat this, but with the chain link mesh (yellow) in the 'Rigid Decorated Group' it all goes haywire !
    0_1486010122066_Dyn2grpRigid.jpg



  • With the chain links (yellow) in the 'Soft' group, the main dynamic loop (red) in 'default' with default settings, and the dangly dynamics (white) in a new dynamic group with modified settings as noted on the image below, here's the latest. Some bits of the dangle are disappearing under her skin,but the dangles seem tobe behaving a bit better. The simulation took about 10 seconds per frame:
    0_1486012811416_ModDanglyDyn.jpg



  • ( today's gripe: I don't like the way this forum changes singlequote underscore default underscore singlequoteto 'default' it's a PITA! )



  • @3dcheapskate I will agree that what ever configuration you had in that last group was more satisfying visually. We also have to remember that these are, at best, estimates as to how items will react in the real world. No way we can account for ALL the real dynamic forces that interact on everything everyday. So yeah, this is Poser. Close enough counts. :)



  • For what it's worth, I'd reduce the friction levels. Sure there will be some friction, but we are talking about a polished metal (low surface friction) that is also very high density (roughly 2.5 times the density of iron). Reduce friction, increase gravity, and the dangly bits should dangle more like the real thing.

    You could also adjust cloth density to actually reflect the values for gold rather than cloth... Cloth density is supposed to reflect the sectional density of the fabric, but given that gold is some 20 to 30 times denser than cloth, I'm not sure how Poser would react setting cloth density to 20, hence my suggestion around gravity and friction.



  • @piersyf - reducing friction seems a good idea. But there's no gravity option in the cloth dynamics (at least not in PP2014) - that's the hair room. I've been playing with the values of various 'Dynamics Controls' and whatever I do there's no real bending within each of the dangles.I think that's what actually looks wrong. So maybe it's the underlying dynamic mesh for the dangles (white) that needs changing? I know that normal dynamic cloth works best with triangles (and I've heard of delauney triangulation) so maybe I need to redothe dynamic part of the mesh for the dangles ?



  • @morkonan - sorry, I didn't see your post until just now !
    Cage's MrLooper scripts are a good recommendation - I'll have to play with them.
    I started looking at dynamic jewellery because none of the non-dynamic ones look right to me. Either they can't be morphed/adjusted to hang realistically for the poses I'm using, or the work involved in doing so is beyond my 'is it worth it?' limit. Reducing the load of the actual dynamic cloth simulation so that it works on one rectangle per chain link was the key - the actual chain link mesh is just dragged along after the dynamics calculations are done.



  • Just got my best result yet with the dangles.
    Different dynamics settings for the underlying mesh for the main loop andthe underlying mesh for the the dangles - non-default values are circled in yellow. I wasreally surprised that setting 'Stretch Resistance to zero for the dangles seemed to give the biggest improvement:
    0_1486136199072_BestYet.jpg

    I also tried with the necklace around her right arm. It's still swinging at frame 90 (so I probably need to adjust the damping), but it's looking reasonable to me:
    0_1486136396163_Arm.jpg



  • Tried a new mesh for the basic dynamic part - a diamond (made from two triangles) representing each chain link, and each diamond in the main loop (white) connected to the next by a single vertex. Dangles are connected in the same manner.This setup seems more intuitively appropriate. I tried three different arrangements for the dangles:

    1. Green -the faces are coplanar with the white diamond from which they dangle.
    2. Blue - the faces are twisted 90 degrees with respect to with the white diamond from which they dangle.
    3. Red - dangling diamonds are twisted 90 degrees with respect to the previous one.

    Here's a Blender screenshot showing only this underlying dynamic mesh:

    0_1486552541398_DynDangles2DynMesh.jpg

    I also created a simplistic mesh to represent the chain links using 16 faces per link, with each link unconnected to any other part of the mesh, assigned to 'Soft Decorated'. Using just a single dynamic group with zero fold/shear/stretch resistanceand zero friction and collisionoffset/depth of 0.1 this is the result after 90frames:

    0_1486553301340_DynDangles2after90.jpg

    No big difference from the gold necklace render in the previous post, and no real difference in the way the red, green and blue dangles seem to behave.



  • @3dcheapskate Vertices from the dangles are penetrating the polygons of the figure, but if you don't mind a little post work, you could fix that with the morph brush.



  • If anybody wants to play, I've uploaded both the latest and the earlier gold versions of the" test necklace with dangly bits" to ShareCG as Poser props (positioned for V4's neck, with all the necessary dynamics saved in the PP2) - https://www.sharecg.com/v/87236/view/11/Poser/TEST-Necklaces-Using-Poser-Dynamic-Cloth



  • @eclark1849 - My own (rather hamfisted) postwork approach is usually to do a spot render of the offending area but with the body part of the figure that is poked through made invisible, and then fix it in GIMP. Your approach sounds more sensible ! :o)
    I was also struck by AmethystPendant's recent post on your mainCloth Room thread,specifically the bit about exporting a the prop as an OBJ partway through the sim, and then reimporting it and doing a new sim so that you lose all the built-up velocity of the poriginalsimthat caused the pokethrough in the first place.
    Another option might be to increase the collision depth/offset.

    I like having options.



  • @3dcheapskate said in Creating necklaces etc for use in Poser's cloth room:

    @eclark1849 - My own (rather hamfisted) postwork approach is usually to do a spot render of the offending area but with the body part of the figure that is poked through made invisible, and then fix it in GIMP. Your approach sounds more sensible ! :o)
    I was also struck by AmethystPendant's recent post on your mainCloth Room thread,specifically the bit about exporting a the prop as an OBJ partway through the sim, and then reimporting it and doing a new sim so that you lose all the built-up velocity of the poriginalsimthat caused the pokethrough in the first place.

    Thank you, but I actually said you DON'T need to export, simply spawn a morph target. :o) This way you don't have to set up all your cloth sim params again



  • @amethystpendant - Even better ! :)



  • Just tried a very simple pendant, using only the simplified dynamic mesh - a diamond made of two triangles for each chain link, plus a bigger triangle for the actualpendant. No modelled chain-link mesh. In the Cloth Room everything goesin the default group with default settings, exceptI made all friction values zero and increased Cloth Density by a factor of 10 to 0.05.Alsocollisionoffset/depth both set to 0.25

    0_1486567962098_pendantmesh.jpg

    After just 9 frames it had draped rather nicely.

    0_1486568097972_pendant.jpg

    After that it bounced back up, wobbled around a bit and got a bit lopsided, and sunk intoVicky's flesh. But the way it dangled still seemed reasonable.

    0_1486568213659_pendantmeshlater.jpg