Mudbox 2009 / Zbrush - Head to Head performance test videos

[janus]

Head to Head

Mudbox 2009 V's Zbrush 3.1




Test #1: Subdivision (or how high can you go)



Watch Online (http://www.dashdotslash.net/Mudbox2k9/head2head1Desktop.m4v)

Apple iPhone version (http://www.dashdotslash.net/Mudbox2k9/head2head1iPhone.m4v)





I thought I'd start doing a few face to face tests between Mud and Zbrush to see how each performs in head to head identical test's. The first one is to see how high a usuable polygon count can be reached in each. HD geometry was discounted as you can't generate maps from it, only usable raw polygons are counted.


Test Aims:

How high can you subdivide?
How fast is rotation
How easily does it sculpt.



I'll do more of these tests as I have time to give both apps an in depth head to head test.

Please feel free to replicate the tests to check results are fair and vaild.


Wayne...





Hardware specs used for all tests: (There is also a 8 gig readyboost drive not on the list)



OPERATING SYSTEM
Name: Microsoft® Windows Vista™ Home Premium
Version: 6.0.6001
Service Pack: 1.0
Architecture: 64-bit
Registered User: Wayne
Language: 1033
Domain Role: 0
Total Physical Memory: 8189 MB
Total Virtual Memory: 8388607 MB

PROCESSOR
Number of Processors: 1
Logical Processors: 4
Manufacturer: GenuineIntel
Name: Intel(R) Core(TM)2 Quad CPU Q9450 @ 2.66GHz
Version: Model 7, Stepping 7
Description: Intel64 Family 6 Model 23 Stepping 7
Address Width: 64
Number of Cores: 4
Max Core Speed: 2666 MHz

VIDEO
Card Name: NVIDIA GeForce 9800 GT
Memory: 1024 MB
Driver Name: nvd3dumx.dll,nvd3dum,nvwgf2umx.dll, nvwgf2um
Driver Version: 7.15.11.7824
Video Mode: 1280 x 1024 x 4294967296 colors

[Kickflipkid687]

Dang... that's pretty cool, thanks. Had no idea Mudbox was that much more capable than Zbrush. I personally found Mudbox alot easier to pickup and use, but everything seems to be pushing Zbrush. But I think I will give Mudbox another go.

[elte]

Is it because Zbrush is still on 32 bit architecture?

Although Mudbox can subDivide till 31 mil polygon, do you really need that amount of polygon to sculpt? And in gaming, especially baking, is all the detail gonna be captured?

Your demo really made me consider to give it another try Wayne..hehe...

cheers.

[janus]

Its partly that but more importantly its due to teh fact that as mud makes use of the GPU and can hand off massive ammounts of processing to your graphics card, and zbrush doesn't (and can't as it makes no use outside of zmapper of your graphics card). As graphics cards are always getting more powerful by the month it gives mud a tremendous ammount of raw power to tap into for performance.

But lets see what results we get for other tests in this series...

Wayne..

[Maph]

Meh, these tests are always biased imho.

First off, you're comparing a 32bit architecture to a 64bit one. Unless you're using Mudbox 2k9 32bit. Which I doubt is the case, otherwise you couldn't reach that polycount.

Second, the way Mudbox works can't be compared to Zbrush imho. Zbrush uses a software rendering system and relies heavily on raw CPU power, whereas Mudbox gets most of it's power out of the GPU.
I can get up to 60mil polygons inside of ZBrush (no HD) with my 8 core and it still wouldn't lag (except when performing an undo operation). But I can't get higher than 42mil (or something around there) in mudbox due to physical RAM limits (4GB FBDIMM atm).
A system with out of this world CPU's and a crappy videocard will yield much better results in Z then in Mud, and a system with crap CPU's and a monster GPU would do the same for Mud.

And third, you're comparing both apps with only one object in the scene. I think a comparison with an actual production model would be all the more interesting. :)
Don't know about you, but I rarely have only one object in the scene unless I'm just doodling. ;)
And performance plays a big factor if you've got several objects subd'd into the millions in the same scene.

Don't get me wrong, I lurv Mudbox; and I wouldn't trade it for the world. But comparing Mud to Z (and vice versa) performance wise is like comparing apples and pears.

[janus]

Unfortunatley Maph people can and probably always will compare Mud to Zbrush. This test was far from a loaded dice...its the simplest test of all a single models subdivision. The whole point of these vids is to give people the raw facts when making a decision..which decisoin they make whether mud or ZBrush is up to them. Although everyone realies that it's not just about polygon counts etc, but many other factors as well.

Doubtless thourhg these tests there wil be some cases mud wins out and in others zbrush.... thats the whole point to find out the facts and not the information either company provides about the others application. I'm trying to prove test any user can replicate themselves to triple check that I'm not loading the dice.

As the current version of 3.1 is all that is available that's what I'm using to test with, along with the many optimisations I have came up with over the years for zbrush.

You've always been a big supporter of mud as I think everyone probably knows. But we may have to agree to disagree on this one. :)

Wayne...

[zeke3d]

both programs are great in my opinion, even though I am a Zbrusher, and have used it/relied/depended on it in more that one game production pipeline and it was the backbone of the work begin done. Most studios have a long way to go before they all start using these 2 amazing apps to improve their work, and then do as you did in your video, for more than just characters and to add more organic forms, detail and quality in to the environments and vehicles as well, some don't even use either program for characters either. Which is why I'm hoping the future zbrush will have a claytools like ability to generate as an option a point mesh that is truly independent of topology with the click of a button, that retains the shape and volume, but ignores possible irregularities that come from long polys, poles and other forms of poor but sometimes, no commonsense way around it geometry. This way more types of models could see a vast ease of use from being imported from traditional 3d programs and then get fancied or gussied up in these poly intensive apps, then the ability to dial the floating polycount to a manageable level for projection mapping, printing... etc. The only problem I had with using mudbox was how companies choose to pick keys that oppose each other as a way to be original or make it to where you have to pick your side, having used zbrush earlier, cause it was out earlier, I did not want to relearn keys, if mudbox had put out a script that made it behave like zbrush I would give it another go, but like in games nothing bugs me more than uniform buttons across the same genres. why cant it go back to the a n b or single red button with a paddle or stick days? Overall thanks for sharing that video, it was quite informative, well done, and a good sell for the app.

[janus]

I'm sure once hte Mudbox SDK is out that someone will have the time to make a plugin that replicates the zbrush hotkeys....who knows maybe a zbrush SDK will do the same for mudbox hotkeys in zbrush one day.

I think eventually it'll all go to point cloud data and then be simply down to user preference etc...

Wayne...

[Maph]

Unfortunatley Maph people can and probably always will compare Mud to Zbrush. This test was far from a loaded dice...its the simplest test of all a single models subdivision. The whole point of these vids is to give people the raw facts when making a decision..which decisoin they make whether mud or ZBrush is up to them. Although everyone realies that it's not just about polygon counts etc, but many other factors as well.

Doubtless thourhg these tests there wil be some cases mud wins out and in others zbrush.... thats the whole point to find out the facts and not the information either company provides about the others application. I'm trying to prove test any user can replicate themselves to triple check that I'm not loading the dice.

As the current version of 3.1 is all that is available that's what I'm using to test with, along with the many optimisations I have came up with over the years for zbrush.

You've always been a big supporter of mud as I think everyone probably knows. But we may have to agree to disagree on this one. :)

Wayne...

Naturally, they're the two main stream sculpting apps out there. As you said, Mud and Z will always be compared, in one way or the other.
But I can't acknowledge the idea of 'raw performance facts' on two different applications that don't work the same at the core level.
In other words, I'm bashing on semantics here, so just kindly ignore my rambling. :)

I didn't know my Mudbox fanboi-ism was that famous? :lol:

[Aftermath]

Let us not forget that Pixologic will put into the app what the industry asks it to. Zbrush is somewhat like an open source application , thus far the pixologic team has done a fine job on trying to make their application better month to month.

Let us not forget that this is an Autodesk application and updates in the past have taken forever to come to light.

Yes zbrush is only 32 bit at the momment and is still only version 3.1 and has been version 3.1 for some time now. They still have released fixes and new features throught the forums and through their site.

I will stick with my Zbrush mainly because they are always checking the forums and interacting with us users trying to find out what needs fixed and what they could add to make our lives easier. I like the support they give to the product i purchased. Some might say im a little biased or a fan boy. I just am someone who cares about what he spends his dollar on.

As far as zbrush's proformance i can get up to 60mil and still navigate fine, and all i have is a quad core and 8 gigs of ram on a 9600gt. Thats alot of verts to push and i think is plenty enough to make nice game models.

[seven]

I'm all for Zbrush and consider myself on that side of the fence... but I remember how long it took to go from 2 to.... 2.5 I mean 3.0-3.1 Pixologic isn't exactly quick IMO. They do put in what the industry asks... but not monthly.

[janus]

Let us not forget that this is an Autodesk application and updates in the past have taken forever to come to light.



Well you may be pleasantly suprised on that score with mud.... a service pack is due very very soon and includes some new features.


Wayne...

[Kickflipkid687]

I like mudbox mostly for it's texturing capabilities from what I've seen in recent vids. Zbrush is definitely more powerful as far as tools go and whatnot, but I like the UI and basic navigation of Mudbox. It really just comes to personal preference.

[elte]

Though comparing each app will never end, I would rather see it as a complement to each other and see them covering what's each of them can't do e.g painting texture in mudbox is like in bodypaint as in zbrush we have polypaint, but zbrush have retopo and transpose which MB doesn't have, zbrush has matcap which is awesome, but MB has cgfx(CMIIW) which is good for game, and so on..

[Aftermath]

Well you may be pleasantly suprised on that score with mud.... a service pack is due very very soon and includes some new features.


Wayne...

Nice, any idea on whats in store for the service pack?

Well while i am still a student i buy all of the software anyways with the discounts and find what works best for my own workflow. Really what it comes down to is the workflow and what will help you speed up your performance. Some people might use different apps for differnt things another app might have like poly paint and what not. Sometimes i work bodypaint -> photoshop -> zbrush(polypaint) so i guess im on both sides of the fence because i like the new AO feature Mud has but im sure if we asked pixologic for the AO to be inserted i think they would work on it. I should be more open to other workflows and not be so stuck on my methods.

Seven: My bad on the wording on the updates i was talking more or less on the updates for tools and what not that pixologic puts out.

[janus]

Nice, any idea on whats in store for the service pack?

Your not going to like this answer, but I know exactly what it contains, but alas am not allowed to say right now. (Yeah I know thats a bit of a bummer.) But I can say there are additions that have been asked for by the user base since release. The ones already in the public domain are the clone brush and addition of AO baking and display improvement for the real time AO. There may well be some very nice new presets.... but I'd better not say anything else before I get shot lol.

Wayne...

[zeke3d]

can't wait to see those new feature videos janus.

[j3st3r]

IMHO this is a very biased comparison (as most of the comparison in any subject). The use of mudbox and ZBrush depends on personal preference a lot. I like ZBrush brushes, and don't like mud's, I like ZBruhs matcaps, and don't like mudbox' materials. And on my old computer ZBrush is way faster than mudbox, and the same was on my computer at work. Even our mudbox fanatic colleague acknowledged it. But it may vary on how many cores you have, are you using SLI cards, etc.
So a comparison like this is pointless, and leads to flamewar

[janus]

I woudn't say they are pointless as many times its the raw polycount you can get that is a major factor in people buying a sculpting app. Illustration of this point is how many moved to ZBrush from mudbox when zbrush had the highest polygon counts.

As I mentioned feel free to replicate this test yourself, thats the whole point. But if there is a chance it'll end up in a flame war then I'll stop this series right here as that is never the intention, just to add reliable information so people can take that into account. There were simply far too many people all over the net passing on imformation such as mud 'lower polygon counts' that were quite simply wrong.

Wayne...