ZBrushCentral

Hardware and Settings Advice?

I keep seeing people’s work and work being done in high detailed mesh without getting much lag. This takes the heart out of me.

Would you guys mind revealing your secrets in your Zbrush settings or hardware with regards to maximizing performance in either or both polypainting and sculpting?

Currently I am running zbrush in Windows 7 x64, all fancy win7 eyecandy disabled. I have 4gig ram (ddr2 800), A GeForce 8800GT, Intel’s QuadCore 6600. All running at stock speeds. This thing starts to lag at 1m points. A human figure dynameshed at 2048 makes this setup chop to probably 15 fps.

I’ve done the much needed stuff like turning off shadows and putting mem at 3g.

What systems are you running Zbrush on and how far can you go in sculpting?

Thank you in advance.

p.s. Not sure if this thread belongs here. My apologies in advance. And if there is already a thread regarding the specifics of what one must do or have in order to Zbrush at max performance, double apologies. But it would be nice if you guys could provide a guide of the sorts dealing with this issue as I believe many would find it valuable.

If you have a 64 bit OS, the first thing you should do is get more RAM. ZBrush is still a 32 bit app but can still use a maximum of 4 Gb on a 64 bit OS. Your OS will need about 2 Gb, so I would suggest you get, at least, another 2 Gb but RAM is not that expensive so get 4 Gb especially if you like to have more than 1 window active at a time. If you have a Q6600 CPU then I’m sure your motherboard can handle at least 8 Gb of RAM. After you upgrade your RAM, in ZBrush, go to ‘Preferences:Mem’ and put ‘Compact Mem’ to 4096 and ‘MaxPolyPerMesh’ default can probably be doubled. ZBrush does not use your video card, so there is no problem there. Also, you can gain a bit more performance by having your Scratch Disk on an SSD.

Thanks Zber2.

If you find any more tips, please do post them.

I loaded the DemoRhino from lightbox and subdivided it 6 times. It would only go to level 5, I had to go into preferences to up the max polycount. Model is at 14.5 Mil. I can spin it in the viewport with no noticable lag or tearing. My specs below.

System Type x64-based PC
Processor Intel® Core™ i5-2400 CPU @ 3.10GHz, 3101 Mhz, 4 Core(s), 4 Logical Processor(s)
Installed Physical Memory (RAM) 8.00 GB

That’s great, Mr. Doug Jones.

If you don’t mind, at what resolution is your test done? Document size as well if you don’t mind.

And do you guys know if it is true that the size of the mesh does affect in some way performance or even Dynamesh algorithm?

Display is 1920 X 1080. Document size was 1322 X 950.

Yes, it does. Ideally, for Dynamesh, your mesh should just fill the ‘Tool > Preview’ window or be a little bit bigger. If you go to ‘Tool > Deformation’ and click ‘Unify’ it will size your mesh to fit the Preview window.

zber2, thank you for this tip and info. This will help a ton.

Doug Jones, thank you for the response. A note though, I too do not get much lag when using Geometry divide with the Rino.

From your experience Sirs, when do you start to notice lag when working with Zbrush?

I don’t really notice any lag until I start getting around 20 million points total and I am on a 32 bit OS with 4 Gb RAM.

The secret to lots of detail in ZBrush is lots of subtools. Always plan your projects this way. If you want really fine detail, divide into as many subtools as you can.

That’s interesting. Do you assign different polygroups to parts of your mesh (while keeping them in one subtool) and does this allow zbrush to run faster?

No, assigning polygroups to subtools does not allow ZBrush to run faster. There are a number of reasons why you would want to assign polygroups…to mention a few…

  1. If you want to quickly hide portions or isolate portions of the subtool by just Ctrl+Shift+click and reversing the selection.
  2. If you want to split a single subtool into multiple subtools.
  3. Adding edge loops.
  4. Inserting half meshes.
  5. For assigning specific UV islands if using UV Master.

…and more.

Thank you, Zber2, for this information.