A 3D workflow with Lars Ivar Stranden
The Lars Ivar Stranden walks us through how he made The Privateer using ZBrush
Taking a character or creature from a quick ZBrush sculpt to a fully realized, production ready asset can be a very obtuse and overwhelming process. During this walkthrough I will try to map out an overview of all the stages in this process, so you get a better understanding of the steps involved.
From sketch to sculpt
To get warmed up for other projects, I often do some quick ZBrush sketch sculpts from my imagination; this guy actually started off as one of these. Using my trusty handful of brushes, I went from two spheres; one for the head and one for the torso, to a semi-final sculpt using DynaMesh. I mostly use the Move and Clay Tubes brushes for overall mass, then DamStandard and Pinch with its modifier setting turned to 100 for detail carving. The hair tentacles were painted in using the Curve Tubes brush, with a little bit of tapering to the thickness along the curve length.
I usually retopologize my meshes when the sculpt is about 90% done, this is because I know that the process of reprojecting the details onto the new geometry will usually introduce a few issues. I often will have to resculpt some of the more troublesome areas such as the corners of the mouth and eyelids. Depending on how much control I need over the edge flow, I'll use either ZRemesher (less control but much faster) or TopoGun (more control but takes way longer) for this.
Laying out UVs
As soon as the retopologized mesh is in and the details are reprojected, it is time to unwrap some delicious UVs. If you're texturing in MARI, UV seams aren't really an issue. You should however plan ahead, so that your UV's are set up to make the texturing phase as efficient as possible. Adding a repeating pattern around the neckline of your character's tunic? Then unwrapping that section into a straight horizontal patch will make it a lot easier to paint than if it's some weird snake like spiral.
Texture painting in MARI Indie
I use MARI for my texture map painting, including tertiary displacement detail (skin pores, fine wrinkles, fabric texture). While the color maps are usually a mix between hand painting and layering photographed, flatly lighted elements, I have started basing my skin displacement detail on skin scans. After purchasing a scan set (usually .tiff files), I color correct them in Photoshop to give me suitable maps to paint from. These tend to work much better than what you can get from just color correcting a standard photo, where things like skin pigmentation can give you flawed shape information.
Render scene setup
This is the stage where I import all of the low resolution meshes into Maya and start testing the displacements. The main challenge here is layering the two sets of displacement maps. Since the maps from ZBrush are 32bit files with no displacement at zero (black), and those from MARI are 8bit files with zero displacement at 0.5, some normalization of values is needed. I do this by shifting the MARI map values down using a HSV Remap node, followed by a Multiply node to control the overall strength. The two map sets are then merged using a +/- Average node.
Taking notes from classical photography
When the displacement is working, I pretty much finalize the lighting. I am a big fan of taking notes from real life photography when planning portrait lighting in particular. Many websites such as Digitalcameraworld.com have great guide charts for classical lighting schemes. Here I used a simple 2-light setup using an Arnold area light as the key light, and another, smaller area light as a combination fill light/rim light. I also added an Ai Skydome light with an HDRi attached, but on very low exposure as I only wanted to break up the Specular hits a little.
The last step is also the most time consuming. Setting up skin shaders in particular is lots of trial and error, and test renders take time. You might also find that you have to go back and tweak some of your texture maps at this point to get the desired result. Keep in mind that Arnold is a physically plausible renderer, so the scale of your model has to be correct for the shaders to behave properly.
After having all of the shaders set up and looking good, the only thing left is to crank up the sampling quality and let Arnold start working away at the final render. For this project I wanted to get as close to final result directly in the beauty render, so what I ended up with is just the plain render with some minor color correction and level adjustments applied to it in Photoshop. The frame was rendered as a linear 32bit .exr file, which gave me all the dynamic range I needed to do whatever color and value adjustments necessary in Photoshop.
Head to Lars's website to see more of his awesome characters
Would you like to share a tutorial? Follow this link for more information
Like 3dtotal on Facebook