You have different approaches for starting your models. What are the advantages to the different approaches?

It depends on the job, but for me what usually works best is to start right in ZBrush and use DynaMesh to create and iterate on a rough sketch. That can later be either finalized in ZBrush or exported in to Max for further tweaking.

I think it will be best if I just describe the workflow I used for both Baird and Shibboleth.

Like I said before, working on Baird I had a previous model available but actually couldn't re-use any parts since they were made in a different style from what we needed. The face was the only element I could just take and sculpt over, bringing back Baird's younger look. I used Clay, Move and the Standard brush for organic modeling.

When I was happy with the head I had to project the result onto something we call the "universal head". To make a long story short universal head is a generic head mesh with pre-prepared UV's that we use to project our custom head onto. This keeps the vertex numbers intact, ensuring that all blend shapes for this generic head work correctly with new custom shape. Those blend shapes are later used for animation and facial expressions.

For this process, projection in ZBrush works just great. It gives nice and clean results, but if you need to tweak some vertices you can simply use Move brushes with symmetry turned on to make the necessary fixes.

What we usually also do is import our blend shapes from Max back into ZBrush onto individual layers in order to sculpt appropriate wrinkles for facial expressions. These are later baked into normal maps and exported to the shader in Unreal Engine.


The "universal head" pipeline was created by Jeremy Ernst to speed up the rigging and skinning process for all the characters.

Going back to modeling, making the armor is always a tricky part of the process. DynaMesh makes it easy, though. It allows you to arrange all the armor plates the way you like, create very complex mechanical parts quickly and re-arrange if needed. That's a huge time saver! Before ZBrush, I had to do everything in Max and it wasn't a fast process at all. ZBrush simply made our lives easier — hard surface modeling became as fluid as organic modeling.

The final step of the modeling process is the "dirt pass" which is basically a detail pass. This is where we add weathering, scratches, nuts, bolts, rust — anything that will add to the final quality and which will show up in nicely in the normal map.

The decimated high poly model is then either exported in to Max, Maya or modo in order to create a low poly mesh. Sometimes we'll do that in ZBrush as well. UV's and map projections are usually done in one of the other apps. For texturing we often reuse our temporary textures created in ZBrush, finalizing everything in Photoshop and checking in the Unreal Engine to see how it will look in-game with the proper shader applied.

After that's done and accepted the model is good to go. Even so, there is always room for improvements and we sometimes go back to something we'd called done in order to change things around or simply make something better.


Sometimes you do the concept directly in ZBrush? What techniques and features do you use for that?

Yeah, doing concepts in ZBrush is just the fastest way to design for me. Usually when we work, we have concept art provided but sometimes we need to change or re-design parts of the model. Other times we even design something from scratch. These latter cases are where we often use ZBrush.

Garron's head was sculpted from zero in ZBrush. Using this workflow I was able to propose many different heads for the art director to choose from, as well as quickly implement his feedback.

Shibolleth's head harness was also designed in ZBrush. Retopology tools were used to convert complex harness shapes in to a cleaner mesh. (We didn't have ZRemesher back then!) The entire harness was later finalized in Max and re-imported into ZBrush for implementation of the "dirt pass".


Bartosz adds:

When 2D concepts are loose and you're not sure what exactly needs to be modeled, it is good practice to spend some extra time doing more detailed concepts in ZBrush. This is especially true when you want your models to have some logical functionality. It is fast and fun using DynaMesh.

If there is only a rough idea of something you need to build, without any directions, DynaMesh is again really handy. Usually in a situation like that we do couple quick DynaMesh sculpts and then choose a direction we want to proceed with.

Sometimes everything looks good in the 2D concept but in 3D it turns out that some parts of the model just don't fit. Doing rough and fast DynaMesh changes gives me a chance to achieve the desired results really fast.

I prefer DynaMesh over the ZSketch tool. For me it's more intuitive and less complicated. With DynaMesh I can create anything I need by just using different mixtures of brushes.

I use decimation to speed up creation of low resolution meshes. I can preserve all the details I need and still have a mesh with reasonable polygon counts that I can work with. For me it is more comfortable and quicker when meshes are decimated, and I can easily export them to Maya or other software for baking maps.

I used the ZSphere retopology tool in order to create low poly meshes. This retopology tool was useful when I needed a nice, clean mesh to work with. For example, when I had a 3D sketch sculpted with DynaMesh and I needed to extract separate chunks from the sketch. I used retopology to create those separate SubTools with nice topology and could later use them to add more details.

My typical workflow is:
I create a rough base mesh in Maya or just start in ZBrush using DynaMesh. Thanks to this feature I can try things out and very quickly see if they work in 3D. I then can just add more pieces to the design if needed and rearrange or quickly apply feedback.

For quick 3dDbases I use Clay Buildup, Move, Slice Curve, DynaMesh and Extract. If shapes and silhouettes are established, I usually do a retopology of assets to clean up the geometry. Then I can project geometry onto cleaner topology and add more detail using alphas from our library (like scratches and withering). Sometimes I add Noise at the beginning just to see how things will appear with some texture added.

When I finish my sculpting, it is time for texturing. In ZBrush I can quickly apply textures and if I want, tweak them in Photoshop. I can also make a low poly mesh based on high poly version using retopology. When I have UV's on the model I can re-bake textures from the high poly version to the low poly one.

Using this method I made Garron Paduk, Assault Barque and Booshka for GoW: Judgment


© 2022 Maxon Computer GMBH. All rights reserved. Maxon is a part of the Nemetschek Group. Terms & Conditions | Privacy Policy
Various patents pending. All other trademarks are the property of their respective owners.