Keep in mind that the realtime cloth physics displayed in the virtual fitting room you linked are actually being run on a remote server. Your computer isn’t performing these calculations itself, just receiving a video feed via browser or smartphone from another computer that is. Whether or not you’re running Blender, those are some intensive processes.
If you’re serious about modeling hi-res figures with realistic cloth physics, you could always shell out for a render farm service like Render Rocket.
@Richard
here is an screen of the wire of my model not to high
- 7.5k character
- 2.3k coat
- 800 poly pants
- 3.4 k shirt
i have made a lots of test with different poly count in the collision object and the cloth mesh
with simple model the simulation is acceptable but wen it came to more detail cloth and model is really slow and this is with static model imagine whit animation i don’t want to think about it…:spin:
with this model i get a nice simulation
Yeah that is my point. You have to understand the workflow. First as I have said proper cloth sims for realistic stuff take time. What you are attempting there is not the way to go about it as I have mentioned in my previous post.
The body count is too high, and the cloth object is too complex. You have to drive that object with another one. That is the workflow even in Maya.
If you really want help with this I can.
Move it to a support thread and I’d be happy to do so.
Let’s not apologize for Blender’s slow cloth sim here. Indeed, there are better workflows, but there is nothing in the polycount of his objects that should be killing cloth sim performance. Blender’s cloth sim just needs some attention and multithreading. Hell, others are even releasing GPU accelerated cloth options at this point. I love Blender, but the habit of telling every artist with a problem that they’re just doing it wrong needs to end. Nothing would ever get fixed that way.
@Richard Culver, speed apart, i don’t understand how you can’t see the different quality in the output, you mention “smarter” workflow but I can’t really say your cloth tutorial example on your site made with Blender, is comparable or even near to Marvelous, Syflex, Qualoth…at all, honestly.
…seriously dude do you think this is a high poly???..everybody who experience cloth modeling ( like the author) knows the workflow for cloths whether your using blender or maya …just admit it it is one of the pitfalls of blender features, one of the many half baked features that need improvement. don’t make it a excuse.
@Richard,
if you want a decent result you can’t use your method of driving the cloth mesh with another simpler, otherwise you miss all the creases that you iactually want, and also a simplified collision body is often not adequate, in fact most of the time I had to use even a further subdivided mesh for extreme poses.
And I’m speaking of posing for stills, to say nothing of animations…
paolo
Take this to the support forums already!
I am not interested in participating in an argument thread. At all!
If you however want to open up a case in the support forums lets go at it with some serious examples of cloth workflow.
I did not come here to make excuses for anything.
But I am more than happy to help with the current solution.
As to Blender cloth I will say this for the 3rd time.
It needs better features and better performance. But it is the lacking features that are holding it back.
I would not mind discussing what those are and why here if it were not for the fact that you guys would not be able to resist turning this into a flame ware between apps.
Not interested.
You have my vote for better Blender cloth. As I said already I think 2 times.
But that is a long way off.
In the mean time the OP asked if he could get help or should be move on from Blender correct?
So I offered help. It still stands.
Or would you rather I just said, nah… use Maya?
Because that is what I use now… if that was not already clear.
not it’s not flame wars between apps… it’s the truth and facts…blender cloth is half baked…your just making excuses
@Richard Culver - i can speak for myself and not really interested in flame war nor fight between apps. It’s a matter of evaluating the output and how to achieve that. What you said sounded like cloth sim is ok, just a matter of adjusting workflow and plus some more new features needed…what i said is that for me it’s not, simply.
More precisely, you say lacking features, but imho it’s the solver itself also, what produce those wrinkles, folds, collisions…just take a plane used as a blanket and let it fall on a sphere, even with a simple setup like this i can honestly see a huge difference between what Blender has, and Marvelous for example (just mentioning Marvelous because i use it, can’t say about Syflex or nCloth), so speed and features apart, is the engine/solver under the hood, from what i can see by using both.
How is he making excuses? He already said everything that needs to be said in regards to blender cloth, and offered to present a workable workflow with current blender features. And also noted that there are better solutions outside blender. I mean really…
Cool yep. I understand.
Thing about Marvelous designer is that it is not in a 3D app and it is not designed for example like nCloth to work with a dozen or more other special effects features and in the case of Blender is not required to be hosted in and work with all the other tools and features in Blender. I am not a programer, but I think we can pretty much eliminate Marvelous designer from the equation as a special case.
Or at least we can agree that regardless a faster cloth solution would be welcome in Blender and Maya as well.
But lets just take the solver. What is that? Actually the solver is not just calculating collisions, self collisions and springs which make the folds, it is also calculating all of the other features in the tool that you can turn on. Like Stiffness, Bounce, Structure and so on. So the solver is actually the entire solution that also uses some kinds of physics calculation.
That in itself could be improved, yes. But also you need to add more things you can tweak. I have not looked lately but there are something like 12 different attributes in Maya nCloth that you can paint weights on.
Then another half dozen or so that are not paintable but do also effect the simulation. Two of the main ones are a push out value and a self trapped check feature. In tandem these things allow you to solve it when the cloth gets trapped which happens a lot with character cloth.
If that happens in Blender you are done. The cloth will likely explode eventually.
Another cool feature is the ability to turn on a visual cue for collision thickness or self collision thickness.
So without these features you can still get decent cloth simulation in Blender. For what it is now with its limitations. But you have to work around a few things and also stick to a good workflow.
Which is also true with nCloth and Syflex. I use the same techniques there as I learned from LightWave and Blender cloth… two half baked solutions…lol
After years of dealing with those two… I was very happy to be using nCloth.
But a caution… sim times are still large and performance is better. But you can not get around a good sensible workflow.
To talk about that more in depth the offer is still open.
I’m curious why you think Marvelous Designer can be eliminated? Digital costume design isn’t my field but certainly every time I hear someone talking about digital tailoring recently it seems to come up. Certainly it is being used in production by big studios both for film and games.
Amen to that, especially in regard to the UI 
Mostly just meant from a consideration as to why Blender can’t be that good. Or nCloth for that matter. It is designed with an entirely different workflow in mind.
Used in pipelines, yes. But not a replacement for a good cloth tool that works within a DCC tool which has to play with others tools and be optimized for a character animation workflow. I think the coding would not cross over to another tool. And likely the reason it is so fast is that it does not have to.
Well from my experience learning how to do things in 3D is 99 percent of the equation. And I think the largest percentage of people using Blender don’t understand this. The smaller percentage of people using Blender come from other disciplines or software and have learn good workflows and discipline in 3D.
You can eliminate the majority of complaints down to two things. 1) They don’t understand the software or 2) they don’t understand the workflow.
After that, there is the hard fact that not all tools are created equal. And unfortunately you have to work hard at the discipline of learning the craft and master the software before you can make a determination as to what are the things that really need to be fixed and why.
It has taken me the better part of 10 years to come to the conclusions I have come to working with LightWave, Blender Softimage and Maya along the way.
Blender is seriously lacking in major features and tools that I need and many larger studios need in production.
At the same time, with the existing tools, and a good discipline you can to a lot with Blender.
I use it here in my studio alongside Maya, Softimage, MotionBuilder, Mudbox and Zbrush. And to date I have trained a half dozen artists formally familiar with Maya or Max in Blender so we could add it to the pipeline and save money on software seats where we do not need Maya and other tools.
And what I spend most my time training here in the studio is not about how to use Blender but how to approach the workflow. That is where 99 percent of my time goes even with graduates of the university.
Artist error or lack of understanding is the #1 problem in production from my experience. Not the use of software.
And second to that is having higher powered tools to tackle the more demanding work. And that is there Maya comes in. And even there, again, it is the approach that is #1. If I left people up to their own devices all I’d get back would be, well this tool does not work. Not good enough. When in fact, the real issue is the workflow or simply not fully understanding the tool. This is based on real production situations that happen in my studio. I am not making this up.
So pointing out that people might be doing something wrong is in fact an appropriate response and will address most of the issues.
For the rest of it (as far as Blender is concerned) it will take a ton of work and resources to add features and fix issues. And I think they are rolling along as well as they can be given what they have to work with. In my eyes the development is moving quite swiftly.
Every time I use Blender again, I have to learn it all over. That’s not because I’ve forgotten what to do, it’s because every time I have to remember every little element of the workflow and which obscure windows/hotkeys I need to use along the way.
This is true of almost every task in the world: Driving a car, painting a house, you name it. If you forget (or don’t remember, I’m not sure what the distinction is that you are trying to make there) how to do it, you will have to learn it again. If you never really learned it to begin with (following a tutorial blindly, or just clicking around erratically) you will have to figure out what you did last time. Be a critical learner and you will waste less of your time.
Is this any different in any other piece of software that performs a nontrivial task? I have used Autodesk Inventor and I can tell you that if you were to just memorize the steps, you would never be proficient with it. You have to get the general idea of the workflow first. Once you have that and run through the workflow often enough, it starts to become natural. But unlike riding a bike, complex computer tasks can be forgotten and if you work in more than one environment, muscle memory can get in the way. If I have written C for a while, I will put semicolons all over my python code.
Writing software or doing 3d are complex tasks, and like in sports, you need a bit of warming up before you get really productive.