Question About Character Art Production

Rigging
1

On current-gen projects, low poly modeling has been the fastest step of the production pipeline. The most straight forward one, while High-Poly sculpting/modeling and texture painting would take usually longer.

With the next-gen specs and some new tools, such as Marvelous Designer that changes how we can make high poly garments, and dDo and Substance Designer, that change how we make our textures, things changed quite a bit.

In some cases now, depending on the asset (weapons and vehicles come to mind) you may not even have a high poly model, since the low poly have enough geo for all the main shading needs, so the normal map can be used just for surface detailing.

Anyways, with the amount of assets that we need to make now for next-gen projects, given all the player customization and simple plain amount of npcs and such, have you guys found any process or tools that speeds up the low poly time?

I ask this because, for instance, right now I’m working on a garment piece for which I’ve used MD for the high poly, and I’ll use SD for textures, giving me much faster results, but when it comes to the low poly, it’s still the same old thing, in fact, slower now because I need to make denser meshes for the new specs.

Of course, I do use a retopologizing software, in my case Topogun, which I really like, but it just feels there should be something faster already. I wished solutions such as ZRemesher could generate more usable meshes, but I can’t ignore all the bad edge-flows that it creates, no matter how many guides I put.

Having the low poly step to be the bottleneck just seems weird. :confused:

Thanks.

2

Simplygon is pretty impressive. I’m not sure it’s impressive enough for a 5 or 6 figure license fee, but it is pretty good compared to anything else I’ve tried

3

…apart from repeated use of the term ‘automagic’, which is unforgivable.

4

Hi theodox, thanks for the reply.

Yeah, Simplygon is sure nice for LODs, but I was hoping to find something that would speed up a bit the production of the main mesh, not just the LODs.

Thanks!

5

We’ve used it a bit for getting the main mesh out of a Zbrush model and in general we’ve been pretty happy. Texture in zbrush at high res, simplygon a reasonably high level mesh, tweak it for animation, auto-uv, then cast the normals and textures from the zbrush model. The key change is getting the model animatable (and ideally, somewhat animated) before UV/textures are applied. That keeps it quicker since there’s less lost work if things change.

6

Interesting, Theodox! Thanks!
I’ll see if I can get more information of this pipeline.

But just so I can try to wrap my head around it a little better, you’ve been able to get reasonable meshes out of your ZBrush files using Simplygon. But where these ZBrush files built using a base mesh that had good topology, or were they the usual Dynamesh type, with no edgeflow at all?
My question here is if Simplygon is able to reduce your ZBrush file because it can understand it’s base topology, or if Simplygon analyses the shape only and deduces a good topology for those volumes. And how much tweaking is it usually necessary to get it adjusted to the animation needs?

Anyways, I’ll try to find more information, and hopefully some videos, of this process.

Thanks again! :slight_smile:

7

We don’t depend on the input topology, but the artists do use remeshing in Zbrush to get a more reasonable base mesh - that does lead to higher quality results.