# Geometry, topology for dynamic items.

• The main advantages for irregular tri's like generated by the Delaunay process are two-fold:
1 - unlike quad meshes tr-meshes have no continuous edge loops or orientation so the mesh as a whole behaves identical in each direction.
2 - tri-polygons are stable within themselves. Quad polygons can collapse, self-intersect, or otherwise show less stable behaviour (unless the math to calculate their response becomes unduly complicated).
In general tri's are more reliable to give a consistent result. Quads give perfect smooth bending when the bending axis is parallel with an edge line, but very bad bending around an axis parallel to the diagonal.
For most 'easy' applications, where radius of curvature is large relative to the edge length, quads can do fine.

• There's a workflow which uses Sculptris (free) to retopologise a mesh with Delaunay triangles - I have some notes on it somewhere, but haven't used it a great deal. I generally get usable results from quads despite all the advice to the contrary. :)

If you find the workflow description I would be interested (just downloaded Sculptris just in case) I tend to use quads as it makes setting cloth zones easier

• I'd also be terribly interested in the Sculptrus workflow.
I'm finally getting to work on a bouffant gown from the 80's with a lo-o-o-ong flowing veil and blusher made from several layers. The light fabric lends itself to this type of geometry....so I've read. Heavier fabrics work fine with conventional quadrangles, again, so I've read.
Thanks for the info.

• Disadvantage using Tri-Meshes with dynamics: A dense polygon-net is required.

I think a well designed Quad-Mesh with corresponding topology is lighter and faster to compute. With equal results in most cases.

• If you find the workflow description I would be interested (just downloaded Sculptris just in case) I tend to use quads as it makes setting cloth zones easier

I'd also be terribly interested in the Sculptrus workflow.

I found the info, which is in a DAZ Forum post dating back to 2010 (i.e. no longer available online AFAIK). I printed it to a PDF, but for courtesy's sake I've asked the original poster if they mind me reproducing it.

If I don't hear back, I'll re-tell it in my own words; but I ought to go through the steps myself to be sure. Sculptris will have changed in the intervening nine years!

Hang in there.

• OK folks, DaremoK3 has given the go-ahead, so here is the Sculptris thread:

http://www.morphography.uk.vu/pdf/CreateDynamicClothWithDynamicTessellation.pdf

In a PM to me, DaremoK3 stressed that the major drawback of this method was "the inability to protect hard edges on open meshes regardless of edge masking". On the other hand, it's free, so as long as you take care not to disturb the edges of the cloth, work on a backup copy, save regularly, wear sunscreen yadda yadda you should be able to get some usable results.

• @englishbob Many thanks, I will have a look through that over the weekend

• That was a blast from the past -- forgot all about my real world illustrations sig-link to my defunct ArtZone gallery...
Thanks again for archiving, englishbob.

Hey all -- This thread is right up my alley...

Since this is for Poser's cloth simulation engine the use of Delaunay unstructured (or as F_Verbaas states -- 'irregular') tris are optimum for folds.

The Sculptris technique will create the necessary tri-structure at whatever size/density you desire, but might have ill effects on open mesh boundaries, non-manifold connected meshes (multiple ruffles sewn to skirt), and edges that should remain hard/sharp (0 - 90 degrees). I did not test for the non-manifold connections, and is only a guess at this time from my previous work nearly a decade ago.

Note, though I am the author of the technique shared, I haven't used it for several years now. Instead I use safer auto-meshing methods through other FREE software including Blender where I have been working on my Marvelous Designer clone for the past couple of years. I have several tessellation algorithms at my disposal for complete cloth creation to re-meshing like the Sculptris technique.

However, one great advantage of the Sculptris technique is the ability to hand-draw (and erase) the tessellation at variable size/densities within the same mesh. I have found no other software which can match this.

Here is example of cloth sim tools I am working on for my clone which are ultimately for my use in Poser or VWD (for Poser/DS/Carrara -- and when I am done, Blender too):

I can find images regarding the variable meshing for Sculptris if anyone needs to see examples -- just let me know...

• @dk3 Oh wow! the Blender implementation is superb, especially as it appears to maintain the edge vertices, I have a script that converted UV islands to Mesh so I could effectively create a pattern to sew together (its very basic and needs work but pairing that with your tessalation script could be awesome!

• I assume that you're 'DaremoK3 over at Renderosity then? Thanks bunches for the information and when you get something usable I hope you put it in the Blender store, I'd gladly through you a few bucks for something that works.

• Talking about a blast from the past.
Before Marvelous Designer came out I was playing around for that same concept. When I learned about the existance of (then) MD2 on offer for some USD120.- I did not think twice and bought it, so that was the end of my search.
From the searching time I remember an algorithm for tri- meshing that allowed you to force local refinement by adding internal lines and specify the edge length of those lines. I considered it would be a useful tool to force refinement where you need it. It is something I still miss in MD. You can add internal lines but you can specify the mesh density only at pattern level.

• @dk3 is that Blender plugin available anywhere, even in beta?

• @amethystpendant
Thanks for your interest in the addon, amethystpendant. It isn't anywhere near ready for a public release, and is a small part of an entire tools tab which is a sub-tab of my MD clone.

That version there is Mark II, but is still only high level macro code with serious need for refinement and refactoring -- just proofs of concept at this stage.

I am currently working on Mark III where I implemented (hackedly) a way to make sure all boundary verts are added to the boundary group, and my original vision for remesh/re-triangulate that auto-invokes the function again instead of manual restart -- but, so far, it is a big hack that relies on doubling the function class at a specific location within the script instead of being able to reference the original -- it works though, just not optimized.

There are lots of other issues I need to tackle, and I only work on it in my spare time when I can.

With all that said, I would love a second pair of eyes such as yours if you are willing to work with me to get it to a simple state I could possibly release to the public -- Is there a PM system in place here at SM? If not we could go email, or I am available for PM at other locations such as Rendo or DAZ...

Regarding the addon itself -- I am not the original author, and the original addon is available for anyone to use -- I just modified the code for my needs such as making it interactive, and adding flags for access to other functions within the DLL.

• @baginski
Yes, I am. Thank you for your interest, and willingness to throw money towards a working solution, but this can not be monetized -- it is strictly open-source licensed, and could only be commercialized by paying original algorithm author a great deal of money.

Regardless, even if I had free reign with the algorithm, I would not create a commercial addon -- I am old school Blender who believes in the old ideal of moving Blender forward for the community - not for ones own pocket -- but let's not go down that (merit on both sides) discussion...

• @fverbaas
You paid \$120 for yours -- man, you got ripped off! The offer was for \$99 and a free, soon-to-come, DAZ Studio plugin.

I paid the \$99 -- have my crippled MD2 (cripple = no permanent license) -- And, am still waiting for my 'FREE' plugin that was promised with the promotion.

Yeah, I have a love/hate relationship with that software/company -- Hence, why I am making a clone out of Blender...

Regarding the tri-meshing algorithim -- I believe I am using the one you were thinking of. It does the internal line meshing, as well as localized refinement -- it even has the ability to create holes in the mesh structure.

The original 2D mesher is named 'Triangle', and the version I am using is the second Blender addon working with it. The first being a bridge system for B2.49 (with working hole system), and this second one utilizing a DLL wrapper that keeps it working inside of Blender itself.

• @dk3 Yes there is a sort of PM system here, but like the rest of the site it has it's quirks (like the close button being hidden!) I've sent you a PM as I would love to try and help if I can be of any use!

• @dk3
MD2 could have been \$99.- also. It was somewhere in that range. My recollection of \$120 could be because of VAT and such. Anyway I had a lot of updates on my MD2 and it still works fine.
They are kind people, but I do understand your love-hate relationship. Do not get me started there.
I do not recall anyting about a DAZ plugin. I made a Poser plugin: MDBridge for Poser., and am working (slowly) on the next release. MD is still getting settled on having a published API, and support is difficult to put it mildly.
The thing is open for anyone who has the courage to hack DAZ code.

Yes, triangle.c. That was it. If you defined the edges counterclockwise it assumed a hole. The algorithm was quite veratile. I looked it up again. Must have been August 2000 I last touched it.

• @amethystpendant
Responded to PM, thank you.

• @f_verbaas
I don't have to deal with VAT, so that might be the variance -- There was most definitely a free plugin attached - it is what got me to purchase with money I didn't even have -- Took me months to recover...

Yep... that's it, but it is now up to version 1.6 (2005) and the holes are automated with use of the PSLG (.poly) and boundary markers.

I love your MD/Poser bridge, but I have neither Poser11 nor MD8, so I can not use. However I downloaded your bridge back in November for study and have been watching your development. I hope to get to your level, and your work could help me accomplish what I want.