ZBrushCentral

should I add things like belts and straps to my retopology?

Hi guys, having a hell of a time getting this done for school on time. I could really use some direction here. (you respond faster than my teacher :slight_smile:

Ok, I have this sleeve I need to retop and uv for game. The base mesh on the left is separate subtools, the sleeve, each strap and each leather arm guard.

At first I started to a zsphere retopology where I started to include all that detail. I’m still pretty new to this so it took a really long time just to get what you see here. I started to get concerned that even I finished, it might be overkill.

So I stepped back and dropped some quick topology brush curves on just the sleeve portion, sub-divided and projected until I got what you see on the right. As you can see a lot of the tiny details are not coming through.

At this point I’m just kinda lost at how I should go about this. Any advice would be huge!

Thanks.

Attachments

sleeve-retop.jpg

Are you planning on baking your normals in Zbrush?
Are you building your meshes to support normal maps without the extra geometry (not using 90° angles against LP geometry)?
Have your tried a bake in another piece of software that is going to use texture density as apposed to mesh density like Zbrush will use?

Personally I would forgo the extra detail geometry, but it all depends on the end result you’re going for. What is your Tri count? What is your texture size? Does your engine support mirrored normals, and if so are you going to use mirrored normals? How much of your character is going to be symmetrical? Is the character going to be rigged, if so what software and what method? Are you going to be using dx9 shaders or dx11 shaders?

Your current lower poly version should work fine, you just need to adjust your projections to get the desired result. Also, subD 8 doesn’t mean anything as it is relative. How many points is that?

I’d leave them as separate islands on the high res model, and a single mesh on the final game model.

Yes, I’m going to use zbrush to bake the normal.

I’m not sure what your second question is talking about. This is one of my final projects for school so it won’t actually be used in anything. I just need to get it retopped, painted and posed. I don’t really have a strict poly budget. I’ll probably use 1024^2 as my texture res. What are mirrored normals? Not that I’m using it but what does the version of directx have to do with taking the shortest path to a lowpoly mesh and a decent normal bake?

I pointed out that I was up to subdiv lvl 8 to show that even that much detail didn’t allow for the project to transfer all of the detail.

It’s definitely going to be a single mesh for game, the question is wether or not to create the extra geo around those arm band details or just somehow create a normal map from the original to be used with the simplified mesh on the right.

Is there a reason you’re going to use Zbrush over Maya or any other baking option?
The second question is talking about the angles of the objects you’re modeling in relation to the surface normals of the lowpoly object.
Mirrored normals are UVs that use a stacked mirror to increase texture density but they cause another problem.
http://wiki.polycount.com/NormalMap/#Mirroring
That should get you all the info you need on normal maps.

dx9 vs dx11 means a ton of differnece. If you are using dx11 you can use dynamic tessellation to create your silhouette changing details if you chose to and then you wouldn’t need to model them in.

No, no real reason I’m using zbrush to bake the normal over Maya. I don’t know enough about the pluses and minus of either. I guess since this character is made up of multiple meshes it might be easier to bake them all in to one map in Maya. shrug

Thanks for the mirroring link, good to know but doubt I’ll be too worried about this character.

My personal preference for situations like this (or any normal baking these days, really) is to use xnormal, or maya/max. All those solutions allow you to bake multiple high-poly parts to a single low-poly mesh. You don’t need to worry about the projection artifacts that can come from reprojecting to a simplified, subdivided mesh. Best of all, xnormal is free.

The extra geometry is overkill unless you are presenting this in a way where you need to see the silhouette of the arm bands. Even then it’s not necessary if you are presenting in something that supports the directx 11 stuff that beta_channel mentioned.