In case you missed it
You might find these articles interesting
I entered the 3D world long ago, back in 2013 when I was 13 and used to simply model stuff in Blender or Max without even know what UVs, materials, or rendering was. I started taking it seriously and not as a hobby when I got into PrimerFrame, where I took a generalist course for 3D animation.
The school, besides having some good teachers, wasn’t the best environment for developing as a 3D artist because of the poor direction and constant disappointment from how the school treated the students and teachers. So after I finished the course, I spent over a year working from sun to sun to compensate for everything the school lacked and then focused on the portfolio. Anyway, I did get some good things from there, it is where I started to play around with ZBrush and character art.
Another good thing back then was the final project; we worked on it with the class as a team:
Then, I worked on a film called Bikes for some time:
And finally, around half a year ago, I started at elite3d where I'm pleased to say I had the opportunity to assist my team with the character production for Activision’s Call of Duty: Modern Warfare.
Getting into Character Art
As mentioned above it was all just a matter of meeting ZBrush and seeing all the possibilities it was giving me.
I’ve been doing characters seriously since 2016, but before that, I did some stuff as a hobby modeling in Max the old way and texturing in Photoshop. In fact, just for the laughs, you can see it here:
If I discover someone better than me in prop art I don’t mind but if I see a character, I want to do better than what I’m seeing. And this is something that happened at PrimerFrame, too, when I watched my classmates and their works.
Piper the Boy: Pre-Production
I started Piper around the time when I joined elite3d and the project was on and off. In fact, those skulls on my portfolio were started later than Piper but finished first.
The project's goal was to go one extra step with my shaders and textures. The concept is from Taylor Lee who was very kind and helpful providing some feedback to me.
When it comes to references I mix a ton of stuff - it means not only real-life references (which are a must) but other artists' work, too, used as goals or benchmarks for my project. It helps a lot seeing how the real stuff is, how more experienced people approach something, and using that information to create your own way.
Sculpting the Body
I see a lot of people creating a full blockout of the character, but what always works personally for me is to work a bit more on the face first in order to get the sense of the character and then translate that to the full blockout. As you can see in the image, I already had the overall character mood dictated by the face before I started blocking out the body.
For stylized anatomy, I usually work as if it was realistic, and then smooth out most of it keeping the bigger volumes and mainly using Dam Standard to emphasize what I consider necessary. That way I make sure it has a sense of construction but with that smooth and “simplified” sense of the stylized sculpt.
The main brushes I use are Move, Clay Build Up - which is my biggest favorite, - and Dam Standard.
Gloves and boots were mainly extracted from the main body, with manually sculpted wrinkles and details. For the pants and the rest of the cloth, I used Marvelous Designer. I am not a pro in MD, in fact, I work really rough and do most of the detailing later on in ZBrush.
Let me take what I call the “tail” (see above) as an example to demonstrate the workflow. I used a Frozen plane to simulate the part of the hip where it is hung to fit it properly in ZBrush later.
For the pattern, I simply made a plane that later on was split into four different sections, so I could use elastic to shrink all the sewed parts.
Once I get elastic on, I usually only play with strength and ratio, nothing fancy or hard to do.
What I want to show with this is that you don’t need to be a pro to use Marvelous - just start with it and slowly progress with things.
I only use alphas when working on the skin. The rest of the details like leather, small wrinkles, or edge wear are hand-sculpted, and fabric pattern is made by surface or directly in Substance Painter.
Going back to the skin, I start by adding noise using surface to give some bumpiness, then another one for some small pore effect, then use some skin alphas. After that, I start to manually add details and breakups to the surface, thinking about where the character lives, how much the sun affects his skin… For instance, if he’s a pirate, he gets a ton of sun damage, the salt from water an all this stuff that may make the skin more or less damaged.
I tried a lot of methods for posing the character like ZBrush ZSphere rigging, Transpose Master, or simply moving geo in any software. However, what pays off the most is actually rigging the character. It might take more time at first, but then it'll way easier to simply go on and make multiple poses, get better geo deformation, etc. Also, there’s no need to make a complex rig.
For the UVs, I split the body into three UDIMs, one for face, one for torso/knee, and one for arms. The rest of the props were managed on different UV sets. On the face, specifically, I like to do some deformation to get way more space on the UVs:
When working with UDIMs you've got to be aware that if you texture in Substance Painter, which as far as I know doesn't allow painting across texture sets, you’ll have to match textures between sets. What works for me when doing it is to create a generic skin texture that I can use to project onto the seams using albedo and roughness to be able to match values.
I did all the UVs in Maya. I used to work in Max but since I tried Maya I did not come back and I don’t think I will for now. It is just much easier and faster to do that in Maya.
To texture different elements and see if they match without the need to open different files, I created one substance scene and placed all the objects in different texture sets, in a way that I could isolate them to only work on one set at a time so that my computer didn't crash. At the same time, I was able to eventually display other sets to see if they work nicely together.
When it comes to texturing the skin, I use a practical VFX workflow. For practical VFX, they use cotton to paint in a way that allows them to simulate some veins, and luckily for us, we have a cotton brush in Painter!
I simply create red, yellow, and blue/grey layers, and then paint with the cotton brush, placing red mostly on "bloody" areas, yellow on "bony" ones, and gray/blue for hairy or desaturated areas. Even though that's the theory, it is a good thing to also blend the colors to get smooth transitions and get an overall skin color.
The most challenging part of the texturing phase for me has always been fabric, I never used to get a nice feel on it. However, that’s not the case in this project because I worked with some scanned sources which I recommend to do.
The thing is, when you look at references there’s also light and gloss info affecting what you see, which makes it so much easier to get something wrong if you don’t have the eye trained. In case of a scanned source, you mainly have just color info (it actually depends on the scan quality, you may have some shadow info as well) which allows you to see an albedo closest to real-life (this is also cool for skin or any other material, not just fabric).
Let me take the pants as an example - the main point was to add the fibers to the albedo, not just one layer with the same intensity all over the model. It was necessary to give them more strength in places like the seam borders that get more wear and shift the color to a more dirty brown on places like the bottom and the back of the legs that might get messy when you sit somewhere.
Filters, generators, and smart masks in Substance Painter are also great tools, but I try to minimize their use because they can sometimes be spotted - yes, you can modify them to minimize that effect, but they are still noticeable. So I use my baked maps as fills for the masks and then break them with grunges or manual painting.
Below, you can see a breakdown of the barrel wood (also, you can get it for free here):
Preparing the Render
I am not an expert in rendering, in fact, I’m still experimenting to find a workflow I’m happy with, but my usual approach is to set a dark color to create a nice contrast with the lightest part and a nice blend with the darker ones which lets me easily create focal points.
For lighting, I usually do a modified 3-point lighting setup, which means one main light, one fill light, and one back/rim light. The modified part comes in adding a top light to create a top-down gradient which is not baked in the albedo color.
If you work in Marmoset Toolbag, I'd recommend activating tessellation, it comes in handy if you have previously baked a heightmap.
Also, use contact refinement, make the shadow resolution higher, enable GI, give some sharpen to the camera to crisp the details, and some chromatic aberration to fake the camera effect.