why not cinema 4d?

[talent103]

What?? Is the rendering multi threaded at least?

[Cutman]

What?? Is the rendering multi threaded at least?

Yes of course but everything else is single threaded apart from the deformers which only arrived in the last release R15!

Rendering is good with the new GI modes in R15 but IMO not a match for VRayForC4D but there's a lot to like about the Physical Renderer especially for shiny NPR mograph work.

Arnold is being developed as some have already said, I use VRay for the physically accurate shaders which is integrated into C4D almost seamlessly. Octane appears to be quite a popular choice too so there's plenty of rendering choice in C4D.

[3D-Django]

First of all, of course the renderer is multi-threaded, and as Srek explains in this thread at c4dcafe, it's code is one of the most optimized for multi-core rendering:
http://www.c4dcafe.com/ipb/topic/77210- ... ntry543902

Let me try to clear some things up, since some things Cutman said are just not true.

While you are right in saying that Cinema 4D can get quite slow with high object counts and isn't multi-threaded in many areas, I have two things to add here:

1. Multi-threading doesn't make sense in every area.
Some processes are simply impossible to split up (or very hard, p.ex. dynamic simulations), and for some it doesn't make any sense because the split-up would take longer than the actual operation. Most modeling operations I think are examples for that.,

2. Maxon is definitely aware of these issues and constantly makes improvements as you see speedups in every version. These might not happen as fast as we users wish, but they definitely happen. As you said, with R15 p.ex., all deformers and the muscle-system became multi-threaded. I don't know why you think there is anyone at Maxon who believes he has "infinite wisdom" and ignore users. The ones I know are very open-minded and incredibly capable.

But I guess most of these optimizations have to be made deep in the core of the application, thus are very time-consuming and can have huge consequences for many parts of the software. Which means they have to be made very carefully and need to be tested well since nobody wants compromises regarding stability.

AFAIK, this is a very common problem of 3D software: parts of the code are simply getting old.
Softimage: The re-write from SI/3D to SI/XSI took many years and SI lost market share to Maya that time.
Lightwave: they tried a rewrite with the "Core" project, we all know where that went.
3ds max: There was this Excalibur(XBR) project, what happened to that? Is it still under development or was it given up?
Maya: I guess they have similar problems
Even Modo, the youngest 3D app, has problems with huge object counts from what I heard.

This seems to be a very difficult task, and I think the way Maxon handles this problem is to gradually renew their code instead of starting from scratch. This is just my personal conclusion of what we've seen in the last releases though, we beta-testers don't know anything regarding long-term roadmaps either.

cheers,
Günter

Newcomer (<20 posts) alert: please use the URL tags - HB

[Cutman]

One of the things I should've said also is that Maxon beta testers will go to any length to defend Maxon, a loyal breed.

Anyone who thinks multithreading isn't always applicable has been supping on the Maxon koolaid for too long or has mistaken this forum for CGTalk. It's like arguing ICE should be single threaded, madness! Xpresso has not seen any sort of meaningful update for years and more's to the point shows no sign of being updated or replaced with a modern nodal programming system a la ICE. There's a noticeable amount of C4D studios beginning to look at Houdini for precisely these reasons.

We get the same old tripe rolled out, "Maxon is aware of these issues" well C4D users are ALL very well aware of these issues yet year after year nothing is done.

For the record, Modo is equally poor if not more so with handling large objects counts.

[3D-Django]

One of the things I should've said also is that Maxon beta testers will go to any length to defend Maxon, a loyal breed.

Anyone who thinks multithreading isn't always applicable has been supping on the Maxon koolaid for too long or has mistaken this forum for CGTalk. It's like arguing ICE should be single threaded, madness! Xpresso has not seen any sort of meaningful update for years and more's to the point shows no sign of being updated or replaced with a modern nodal programming system a la ICE. There's a noticeable amount of C4D studios beginning to look at Houdini for precisely these reasons.

We get the same old tripe rolled out, "Maxon is aware of these issues" well C4D users are ALL very well aware of these issues yet year after year nothing is done.

For the record, Modo is equally poor if not more so with handling large objects counts.

Please read my post carefully.
Nowhere did I state that XPresso or ICE should be single threaded. The example I gave was modeling operations. And yes, multi-threading doesn't always make sense.
Of course, for XPresso it would, and many other areas as well, no one is denying that.

I tried to explain the status quo of Cinema4D's development and why some things may take longer than you'd expect as obejctive as possible, and I invite you to stay objective too.
No need to feel offended.

Günter

[noseman]

As a matter of fact, although Multithreading IS always possible (programmatically), that doesn't mean it will yield computational benefits in many cases.

The easiest example is the domino effect.
A single row of dominoes, for example 24 of them, each assigned to a single thread in a 24 core system, won't scale 24 times, because for each consecutive domino, it will fall only when the previous one touches it. Therefore, although there are 24 processes running, they will execute linearly. This is one of many algorithmic cases where multithreading doesn't benefit the task at hand.
If you had 24 non-interacting domino simulations running at the same time in the same scene, providing one doesn't influence the other, then in theory you would have benefits. But that is a rare case. Would you rather have Maxon divert development resources to address that, or focus on something more useful? I, personally, choose the latter.

XPresso is single threaded, nobody said otherwise, and Maxon is surely working on better implementations of modern technologies to address all shortcomings, as I'm sure all 3D software companies are, as well.
This weird concept, some people have, that Maxon "doesn't care", is contrary to the fundamental principle of creating software. I would rather you said, "Maxon is not fixing what I want", and then I may even agree with you in certain areas. Only, I'm very well aware that Maxon is not creating software only for me.

Although I confess I am a huge fanboy of Cinema 4D, I would never defend the un-defendable. What Maxon beta testers ARE doing, is working hard, with Maxon, so that ALL users get a better product, and a wider choice of great 3D software.
To prove my point I have to state that I have invested many hours creating free videos for no reason, other than to break down certain preconceptions about Cinema 4D. I will not directly benefit monetarily or otherwise, unless you consider positive feedback to be a benefit…. Now that I think of it, it does make me feel better about myself.

To finish this off, I apologise if I sound harsh, but you can apply everything you said to any company and any product, because it is a generalisation, not a fact based assessment.

As a gesture of good will, I will offer you, as I have offered to many others in the past, if you have specific workflow issues with C4D, I would be glad to assist you if it's within my technical abilities.

[noseman]

ooops, two fanboys replied…

Just kidding. We are all in good spirit here!

[Srek]

Maxon Betateser tend to keep critique in the family. Anyone who has seen what happens in the beta area of Maxon quickly recognizes that they are not cutting us at Maxon any slack.

[Cutman]

As a matter of fact, although Multithreading IS always possible (programmatically), that doesn't mean it will yield computational benefits in many cases.

The easiest example is the domino effect.
A single row of dominoes, for example 24 of them, each assigned to a single thread in a 24 core system, won't scale 24 times, because for each consecutive domino, it will fall only when the previous one touches it. Therefore, although there are 24 processes running, they will execute linearly. This is one of many algorithmic cases where multithreading doesn't benefit the task at hand.
If you had 24 non-interacting domino simulations running at the same time in the same scene, providing one doesn't influence the other, then in theory you would have benefits. But that is a rare case. Would you rather have Maxon divert development resources to address that, or focus on something more useful? I, personally, choose the latter.



As a gesture of good will, I will offer you, as I have offered to many others in the past, if you have specific workflow issues with C4D, I would be glad to assist you if it's within my technical abilities.

Have you used XSI and in particular ICE at all? I've read plenty of Maxon cases against multithreading to the point I'm immune to them.

I'll take you up on your offer,

Perhaps you'd recreate the dandelion scene and show similar viewport speed in C4D with the number of dynamic objects and mesh densities as is being demonstrated in Softimage.



Then come back and justify single threading over multithreading.

I don't think it is particularly helpful to those that might be considering migration from XSI to C4D to try and brush over the serious shortcomings of C4D performance, do you think they won't notice when they download the demo? What's the point of a migration thread if you can't be honest about certain aspects of the application being considered?

The Cinema 4D workflows for doing heavy mograph work are to lores proxy everything, solo layers or to constantly do viewport preview movies whereas in XSI with my modest experience you're able to work closer to final output for longer.

I think Softimage continues to make a bloody good case for itself despite being EOLed and is many years ahead in performance of C4D. There's no certainty that C4D will ever attain this level of performance, check out some of the other ICE demos of crowd effects, the performance is in a completely different ballpark.

Like I said before C4D will be attractive to those XSI users that aren't all that big on ICE or are TDs that want a fully featured application that does most things well enough and being developed. Those heavy ICE users and TDs will probably find a spiritual home in Houdini. If C4D had the performance to deal with the scenes normally undertaken with Softimage it would be a no brainer for many more migrants.

[noseman]

This is not a VS thread, nor has anyone lied about Cinema.
I have already regretted replying to you in the first place, but anyway...

As for the scene, if you are kind enough to send me the models, I will honestly give it a try using Cinema 4D and although I'm sure Softimage has a faster viewport, I'm sure setting it up is faster in Cinema 4D and mograph.

Awaiting the scene models.

[Mathaeus]

Then come back and justify single threading over multithreading.

Well a bit of nit picking. I also have impression that ICE is so much stronger and faster than C4d Mograph or xPresso, for about factor of 10 or more. And that is not motivating to go into c4d, deeply - nicely to say.
But, I don't believe it's just about multi threading. There are many ways to make such thing to work faster, like splitting data into chunks, removing any data that doesn't contribute in final result, a way of automatic caching under the hood, SIMD instructions, what's not.
In short, if you're an artist ( like me), all what you need to say, is what you already did in previous post - to show example.

[noseman]

Since I put my money where my mouth is, here's a quick demo.
Single threaded and all.

https://vimeo.com/89718561

[Bullit]

Looks good.

What you think will be next steps of development for Cinema. I know they can't say but what you feel they will be noseman?

[noseman]

Nothing I can say, although even beta testers are aware of a fraction of what's going on inside Maxon.
One thing I can say, is that the Dev team is listening, and their listening very carefully.
Creating such a stable, easy to use program, from a company that's not going anywhere anytime soon, that has been growing fast but steadily, can only be achieved if they are very careful.
Of course, I, as well as many others, are curious, but I have happily surrendered that for the reasons mentioned above.
Hang in there, with us, wait, and see. Exciting times are coming

And another video showing the piston scene using 100% Dynamics (as usual it's only single threaded):

https://vimeo.com/89724411

[MauricioPC]

Hey noseman ... that's some awesome videos. I really liked it, keep up since I'm loving your videos. lol.

[luchifer]

Hi noseman, I dont want to abuse of your good will, but I have a c4d question (wich of course, involves XSI workflow). I ask this because im unable to find a workaround:

Is it posible in c4d to create a curve, sweep it to make it a tube, extrude some polygons to make a nice looking tube) and then change the curve initial shape? (as you know, in XSI you can change the points of a curve after you extrude it, but in c4d I lose the ability to modify the shape once I make it editable)

I saw in modo the workflow of adding shapes to tubes or parts of a model, and this can be done easily with mograph and selection tags, but in order to add pieces to the tube, I have to make it editable, so im losing the ability to change the shape of my curve.