1027x764

1606708376640.png

๐Ÿงต Untitled Thread

Anonymous No. 845855

I am learning cloth animation in industry standard programs. What is the recommended workflow?
>You simulate t-pose in Marvelous for 3 seconds to get it to settle and then you send it through vellum or nCloth?

If you do this, don't you lose the advanced cloth simulation features of Marvelous but gain the ability to react to regular Houdini objects / forces? Isnt vellum a lot worse?

Anonymous No. 845859

i can't answer your question, but i'm curious

>don't you lose the advanced cloth simulation features of Marvelous

what can you do in marvellous that you can't in vellum?

Anonymous No. 845860

>>845859
>what can you do in marvellous that you can't in vellum?

>stitch the clothes and create the pattern
>have the seams affect the solver in a much more complex way than if you did the same with vellum, according to odforce kiddies

Anonymous No. 845867

>>845860
authoring patterns in houdini would probably be a pita.
guessing it's possible to stitch a premade pattern inside houdini - probably a bunch of manual grouping work tho.

seams, no idea and not sure what you'd be looking for in particular. maybe vary constraint properties around the seams?

Anonymous No. 845869

>>845867
also, sorry i couldn't be more help; to make up for it here's the tutorial that your screenshot is from:
https://www.bilibili.com/video/BV1uc411h79R?p=1

Anonymous No. 845873

>>845855

Nothing comes close to Marvelous Designer regarding cloth animation. It's pointless to even try other software.

The issue is with processing power. On an i7 I could simulate at 1 fps a >200k polygon clothing. On the Ryzen 5950x processor I can simulate the same scene at a whopping 5 fps. GPU simulation in MD is only useful for asset production but it is much much faster. The company however said they had no plans on fixing the GPU simulation.

Cloth simulation is something I try to keep up with regularly. All the papers that are coming out are still useless for actual production. For instance they can simulate a really complex piece of clothing but you can't manipulate it in real time. Or you can't simulate several layers of clothing at once. Or they can't simulate several layers of clothing at once along with hard surface elements like a zipper for instance. Marvelous Designer can pretty much do anything you want but the processing power cost is far too great.

The only viable solution now is for MD to dive into post-production development rather than asset development. The ability to fully tweak the physics engine will give the average Joe Pixar-level results even with working at 5 fps.

Anonymous No. 845874

>>845873
>Nothing comes close to Marvelous Designer regarding cloth animation. It's pointless to even try other software.

so how do you react to other objects in the scene if you're doing VFX? Like if i have the woman in the dress and she's getting shot at with bullets and you need them to tear the cloth and then have bleeding affect the textures and have the blood pour out and effect everything and then have the lady dynamically crumple to the ground. And then you would have to keep iterating on this shot over and over.

Anonymous No. 845889

>>845874
>i have the woman in the dress and she's getting shot at with bullets and you need them to tear the cloth and then have bleeding affect the textures

First off: wtf.

Second: all of that can be done in Marvelous Designer.
- the dress can be an MD patterned asset and in this case it's all pretty straight forward if you ever used MD
- if the dress is in the form of an external OBJ have it imported as cloth from obj file in MD
- import the animation in the MDD file format including the bullets
- run the simulation
- export the simulation as MDD
- DONE

To achieve cloth penetration and ripping is all in the modelling, not the cloth simulation. There are several ways to go about this. What I would do is add pre-tears in the OBJ where the bullet hits. These tears can be linked together using modelling tricks like using several extra edges linked to the same vertexes. This trick does not change the vertex order but tells the simulation engine what to hold together even if not physically linked.

In the final animation you can add an extra OBJ geograft that attaches to the area where the bullet hits. This new OBJ will get all the extra textures like blood or fine ripping effect.

>then have the lady dynamically crumple to the ground. And then you would have to keep iterating on this shot over and over.

I don't think you're aware of the MDD file format. MDD is just a series of morphs meaning you can import any type of animation you want into MD or any other software which supports MDD.

Anonymous No. 845892

>>845889
just curious why are you advocating importing / exporting in MDD instead of alembic?

Anonymous No. 845894

>>845892
>alembic
MDD is pretty straight forward.
Alembic is more complex and I had issues with exporting/importing alembic from certain packages. Alembic works too for the workflow I described if you have a workable pipeline.

Anonymous No. 845897

>>845874
just use houdini

560x769

59418-large.jpg

Anonymous No. 845899

>>845897
try animating this in Houdini

Anonymous No. 845902

>>845899
just change the script to present day. ez.

800x600

s.jpg

Anonymous No. 845903

>>845899
How would one animate this in houdini?

Anonymous No. 845913

>>845903
with mocapped data it would be the same as anywhere else, scrub.

Anonymous No. 845916

>>845913
lol
we're talking about cloth animation here

>imagine having to do mocap on actual cloth

Anonymous No. 845919

>>845899
Why wouldn't you be able to simulate that in Houdini? Houdini is definitely not good for designing, but once you have the mesh exported/retopoed from MD, you can do whatever you want with it.

Anonymous No. 845963

>>845919
Too many layers. There is not a single AAA studio who animated something even remotely similar and you're sayin it can be done...
Pixar had major issues with their proprietary physics engine when making Coco.

Anonymous No. 845964

>>845963
The average AAA wagie is a brainlet and the ones who could build it are smart enough not to do it on a salary for some jew.

Anonymous No. 845965

>>845889
I'm not sure what an obj geograft is. And what's adding extra edges linked to vertices? I don't think you know what you're talking about.

Anonymous No. 846007

>>845965
the only people talking about geografts are dazlets.

Anonymous No. 846008

>>845964
Well yeah, I can animate that piece of clothing no problem. I also have no intention in helping AAA showing them how.

>>845965

>geograft
geometry graft
based on vertex position one obj is grafted on another obj flawlessly
video game character have the body and head separated in 2 separate files, combining these together into one mesh is called geografting one mesh onto the other. It's old tech, probably everyone calls it differently based on what software they use.

>adding extra edges linked to vertices?
Physics sims run by simulating vertices. It knows what vertices are next to what based on connected edges. Adding extra edges can link vertices that are not linked together in the actual OBJ without ruining the vertex count. With this method you can easily animate a zipper going up and down.

> I don't think you know what you're talking about

I don't think you've got the know-how to understand what I'm talking about. Study first, judge later.XXTMW

Anonymous No. 846031

>>846007
Coomer fag is in here talking about physics sims simulating vertices and video game characters coming in two OBJ files. Is this what Chads felt like when I showed up to parties in college?

Anonymous No. 846043

>>845899
desu considering a garment like that would have a sturdy crinoline underneath and would move as one shape, it might actually be easier to animate convincingly than OPs pic

Anonymous No. 846094

>>846043
what about the hat, umbrella, and sleeves?
yeah, that's what I thought

Anonymous No. 846095

>>846094
>what about the hat, umbrella, and sleeves?
What about that? Just animate them as well.

Anonymous No. 846116

>>846094
i guess? i'm not the guy youre arguing with

Anonymous No. 846216

>>845889
You'd never use MD for something so dynamic because you're gonna need extra velocity ripples and the likes for extra drama that are gonna be generated from manipulating the bullet hits. MD is hands down the best for garment modelling, and a joy to use for that purpose compared to vellum stitching. But it can't do robust automated iterations the way Houdini can, nor can it do beautiful volume driven wind effects.
>>845965
anon does get it. Constraints are everything in sims
>>845903
https://youtu.be/Mi7hWOubIFY?t=1145

Anonymous No. 846228

>>846216
>Constraints are everything in sims
no shit. he just doesn't know enough to call it a constraint, or that md was built for tailors not cfx, or that volume advection and iterating are a thing.

Anonymous No. 846236

Is there an industry that wants cloth physics? Movies maybe but i doubt games would need advanced cloth physics

700x700

DpQ9YJl.jpg

Anonymous No. 846247

>>846228
>geometry grafts
Ackshyually it's "constraints..."

Anonymous No. 846248

>>846228
You didn't even know it was called a constraint or geograft before others pointed it out to you lol

Anonymous No. 846249

>>846216
MD has a bunch of limitation but those are because of asset creation purpose. The physics sim is the best I've worked with. It's a shame not to use it.

Look at this problem Pixar had:
https://youtu.be/U8U2xyyEhjs?t=43

The MD engine can deal with that no problem.

Anonymous No. 846256

>>846247
not that guy but i don't think anyone's saying 'geometry grafts' (lol) are constraints.

Anonymous No. 846264

>>846247
>>846256
"Geograft" is a term in DAZ. Kinda like a rivet, I think.

Anonymous No. 846270

>>846264
daz guy's explanations of marvellous' sim capabilities just leave me confused t b h.

1006x560

1627688329456.webm

Anonymous No. 846490

>>846270
I choose to ignore whatever Daz users say. Shit like:

>>845889
>What I would do is add pre-tears in the OBJ where the bullet hits.

Anonymous No. 846499

>>846490
the guy literally mentions pre-tears in how he made that scene lmao

Anonymous No. 846507

>>846499
He painted a mask to art direct the tearing, but the tearing itself is procedurally driven. He mentions pre-fracturing, only to discard it as it would have looked worse.

Anonymous No. 846532

>>846507
like I've said: he literally mentions pre-tears

there are n ways of doing things
he could have replaced the figure with the pre-teared obj at the exact moment he wanted the tears to appear

essentially you run 2 simulations for the cloth, one until just before the tearing stops, take this morphed obj with the simulation baked-in, add tears, place it in the scene, continue the simulation.

You can add as many pre-tear instances as you wish. With this method the artist can control the tears manually. This is also a low-tech solution because there has to be some packages out there that can split vertices while simulating.

Anonymous No. 846533

>>846490
>>846499
>>846507
technically he fractures every single tri/quad on the mesh and lets the stress * mask handle where the tears happen.

i'd never noticed that if you've got a vellum cloth constraints node, it won't create bend constraints across fractured pieces

the weld constraint's description in help is weird, but it makes a lot of sense when you have an extreme example like this:

if you have the 'add bend across welds' enabled, it basically ends up creating the missing bend constraints and you just end up having a standard cloth set up again.

the only difference is just some new attributes and the all important break threshold.

i'ts very neat.


>>846532
nah, you're completely wrong here, dazguy.

Anonymous No. 846534

>>846533
also, for clarity, as insane as this 'fracture everything' set up seems, because not much is different b/w it and a regular cloth setup, sim times are comparable (i think).