How To Enhance Hand-Painting With Blender Shaders For Stylized Character Art

Introduction

Hi there! My name is Cherylynn Lima, and I got into 3D art about 9 years ago back in college. Around then, my goal was to be an illustrator/concept artist, but I found a love for 3D in the low-poly community, and my career started rolling when I made my first YouTube tutorial channel. My years of studying fine art and a passion for tech combined well in a field that requires a lot of learning and application. I also had the good fortune to fall into 3D right around the time really art-heavy indie games were booming (circa 2016) and this had a huge impact on my love of 3D, characters, and games.

Does anyone recognize this old tutorial? If you watched it, give me a shout-out. I love hearing from my old subscribers. Eight years ago… oof!

A couple of example pieces from my illustration background. I find having drawing/painting experience immensely helps with stylization. Hope you can feel the effect on the end result of my model!

In terms of projects I have contributed to, on the stylized indie front, I have created character art for YouTube giants like Markiplier (Damien, 2019) and Game Grumps (Homebody, 2023), contributed art direction to BAFTA-winning indie darlings like Before Your Eyes (2019), among a plethora of others. In the AAA environment, I made character art for The Last of Us 2, MLB The Show 20, and, most prominently, I was a Character Artist at Bungie for 5 years (Destiny 2/Marathon).

I am constantly learning and honing my craft, whether that be through painting, artistic and narrative exploration, or expanding my technical craftwork. My interests stray from avant-garde, super artsy indie games to stylized classics like Legend of Zelda to epic worlds like The Witcher 3, Dark Souls, etc. My style tends to shift with the day, but at the core, I love light, color, and story. I try to learn and bolster my traditional art skills with tech art whenever possible. Thanks, YouTube!

The Black Gold Girl Project

I worked as an armor artist at Bungie for the past 5 years – and during that time, I learned a lot about detailed fantasy/sci-fi/mech-stylized realism. I became skilled at defining material reads. I also developed efficient strategies for working quickly and cleanly and learned my way around shader graphs. But I felt like I was missing something: the colorful, stylized, hand-painted styles I used to art direct during my indie days. So, I decided to make a personal piece that blended my fine art skills and the expanded technical and visual experience I'd gained working in AAA. This culminated in Black Gold Girl.

Black Gold Girl was originally a concept piece by the uber-talented Wu Shenyou:

I was mostly drawn to the thick, tangible paint strokes in the model. You can really feel the form language with how he chose to highlight and shadow his art. I was especially drawn to the polygonal, exaggerated shapes of the light bouncing off the reflective bodysuit and to the chunky, stylized shapes of her jacket and hair. I wanted to imitate the original drawing as closely as I could and make something that felt fresh, appealing, and organic.

Inspired by games like Fortnite, Overwatch, and League of Legends (Arcane, am I right?), my main goal was to create an AAA-level stylized/hand-painted piece. At the beginning, I wasn't sure what that meant, so I took some time to research the stylized greats before me. I paid extra attention to how they rendered cloth, simplified shapes, and used hand-painting to enhance lighting and form reading.

I mused upon how to achieve this effect in a way that felt fresh and distinct from the AAA giants. Once I'd grasped the visual style I was aiming for, I felt ready to tackle the model. 

Modeling

I almost always approach a model the same way, and that is to spend as much time as possible in Blender. Whoa! You might say. What about ZBrush? To that, my dear reader, I'd reply: the more time I spend in Blender, the better for my mental health. This program is a godsend because of its unique, non-destructive workflows. This means that I can stack modifiers and undo things to my heart's content, and I can use those modifiers to duplicate and instant and modify shapes in ways that permutate across the model. I can also be very particular with my edges and forms, pushing and pulling vertices without destructively adjusting the shape. In a lot of ways, you can say my technique is inspired by the subdivision modeling workflows of older-era Pixar (think using Maya to model Woody from Toy Story before ZBrush existed).

This is a wireframe shot of the model's face. I usually work with a mirror modifier until I need to unwrap the surface. I keep two windows open, one with my working model (right) and one with my model with overlays (wireframes, etc.) turned off (left); this way, I can get the cleanest read on my forms as I push and pull vertices.

I use creasing (purple lines) to get really tight folds in the geometry, and I am careful about where I have tris and quads as it causes odd shapes on the surface. I polymodel some pieces (eyelashes) but paint others (eyebrows) because I'm thinking ahead to how these things will animate and what will have the most effect on silhouette. This is the general way I created all the forms on this model, starting with simple basic planes, adding edge loops and carefully increasing complexity, crafting the base shape of the underlying geo, and then adding subdivision levels until I got a juicy result. It's a cleaner way to work with stylized forms, it keeps it simple. This also makes unwrapping very simple later on.

The full body follows this same logic. I box-modeled all of it from mostly a single plane:

Polymodeled clothing wrinkles are easier for me to mentally envision and control:

Even the hard surface gun was created using this method I often used at Bungie (much to my coworkers' surprise): basic mirrored shapes like cylinders and boxes, slowly building up poly modeling complexity, working from large forms to small forms. I used a couple of booleans and arrays to help keep the form clean. The topology is actually very simple and approachable if you take it one small piece at a time.

I also poly-modeled the hair. The braids were easy: I just modeled (and unwrapped) one piece and then used Alt + D (special instance) to duplicate that braid around the head. Special instancing in Blender allows one object to adjust all the others the same way. So, using that one braid as the "parent", I then posed the "children" braids with curves (using curve modifiers). This way, when I altered the first braid, all the other braids had the same adjustments but could be different arrangements.

Using a curve and curve modifier to control the instanced braid:

Topology & Shading

The texturing, hand-painting, and material reading of this model were going to be essential for the final result. So, taking into account my references, I devised an experimental technique for this model using a mixture of hand-painting, PBR materials, and anisotropy.

I actually recorded a whole video tutorial walking through this step by step in detail but I'll break down my process here.

For this model, I wanted a shine to bounce off the forms of this model and make it pop. So, I did some research into how to achieve this and decided to use anisotropy. My understanding of it was its applications for hair cards (think of it as the hair shines on an anime character: reflection follows the form in a defined direction). I researched how to add this effect using shaders in Blender.

In a nutshell, I created Fresnel (Eevee doesn't allow fresnel, so you have to fake it) by using a Fresnel node, driven by the UV map's direction, then made bands of color by breaking the values with the Color Ramp, then plugged this into Base Color. You can learn more about this confusing shader trickery in my video tutorial.

Important note: The anisotropic effect is controlled by the direction of your UVs. If you make them horizontal, vertical, or diagonal, that will all change the way the shine glances over your model. Think of the red line being which direction the shine will ripple over the surface. You can use a map instead, but I just oriented them manually how I liked.

GIF of early shine testing on bodysuit without hand-painted texture:

The body UV map (unwrapped in Bblender), with the red line showing the direction of the shine:

The same effect is shown more clearly here on the hair, using strips of geometry unwrapped in a long vertical rectangle:

UV map of each hair card (individually unwrapped for hand-painting purposes). They all face the same direction for the light to glance the same way.

Texturing

With that shader set up on all our materials, I started hand-painting. As mentioned, the stylized hand-painted effect you're going for will change dramatically depending on what lighting scenario you pick. Your model can be almost entirely lit by the lights in the scene (like here, in Fortnite). But I feel this can look a little flat. Where's the painterly juice?

In Arcane, nearly 100% painting can look amazing, but you need a lot of skill to make it read, and each lighting scenario may look very similar (baked light from the top).

Once you've decided on how much lighting influence you want, you can set your scene up. I went halfway between painted and real light, so I focused on defining some forms with sharp painted highlights (as indicated in the concept drawing). However, I didn't fully define the model's lighting with painting; I used my own lighting setup.

The model only showing painting without lighting:

Boom, lights on:

This is a little difficult to describe without going into painting and color theory at large, but the biggest takeaway is that you want to think about creating fake lighting on the model without distracting from the main lighting in your scene.

So, for me, I start by hand-painting the AO of the model. The AO (ambient occlusion) is all the areas where light can't quite reach: think crevices, wrinkles, folds, etc. You can always start your hand-painting there, with the aim to enhance the lights you will add later.

In red, I've marked areas of super common AO or areas that are almost always in shadow. The under eye because of the brow casting a shadow, and folds in the skin like the eyelids, corner of the mouth, and nostril.

See? Arcane did it too:

Then, I focused on building up highlights on areas that protrude more. 

In green, I've highlighted the areas I think should receive the most light (the concept helped inform this as well, but the idea is the same). I gently brushed a lighter color onto these areas.

From there, I used mid-tones to define the transitions between them. In some areas, like the chest here, I really pushed the distinct highlight shape from the drawing since I liked the way it looked.

Here's that painting, enhanced by an added anisotropic effect and lighting on top.

This is the basic idea for the entire model: Focus on defining the areas of concentrated shadow, then define the lit forms. You can use really stylish shiny shapes to push form further.

Another example: I hard-painted this shine on the sleeve to really make the form and edges of the folds pop, using dark tones to make the wrinkles look deep.

As for the lighting and rendering, I used a setup like so:

I used sun lights (red) to define the core lighting and shadows for the model and area lights (green) to define sharp rim lights that appeared in the concept and highlighted the edges of the form.

Rendered in Eevee's viewport shading:

Rigging & Animation

So, about midway through the modeling, I got some feedback from my mentor, Satoshi Arakawa, that I should try to pose the model to capture the main forms from the image more closely. I figured I could retain my mirror modifier by giving my model a rig and posing her so I could switch between her A pose and her concept pose. This allowed me to really enhance the forms and give her personality. So, I keyframed the concept pose and the original A-frame pose and switched between them as I worked. The two sleeves are the outlier here, as I couldn't quite match the concept's forms by rigging alone, and I opted to model them separately.

This had the bonus effect of… rigging the model! I was able to animate the model quite easily from this point, as I had already weighted everything to get her into the pose. I strung together a couple of looping idle animations and combined them with the climactic POP of the shooting animation. I timed a bonus camera bounce for effect.

The only additional things I added were some bones to her hair, earrings, and tassels, which I gave some physics sim using Blender's Wiggle Bones add-on. This basically adds simulated bounces to the original animation.

I parented these additional hair bones to the main armature head bone so that they would naturally follow the body as it animated. 

Warning! This is not a game-ready rig, this was a fun flourish I added to a rendered model. Each bone substantially increases the load on the engine, so adding this many bones will be an eyesore for tech artists.

For the gunshot, I created some VFX planes with a scrolling shader effect and had them in position to be visible at the precise moment she shot.

And then I rendered it all through a camera. It's just that simple!

Conclusion

All in all, this model took about a month to complete. I had a large portion of the base forms modeled in a couple of days, but figuring out my shader workflow was a lot of trial and error. Additionally, adding animation, VFX, and producing a tutorial added about a week (but it was worth it and was a fun challenge!).

The main challenge I faced was figuring out how to combine my knowledge of shaders to create a cohesive final effect. I am more experienced with Unreal Engine's shaders, so learning Blender's nodes was a new challenge. Learning how to use physics simulation with bones was also a new challenge, but thankfully, Wiggle Bones made it really easy. 

My advice to beginning artists would be: YOUTUBE. YouTube. Watch every video you can. You can build your knowledge base so quickly with those tutorials. Find masters of their craft show you how to create things you're interested in, copy them, and try to break down what you see. Analyze as much as you can, try to watch what you're seeing actively, and figure out how you can change one thing about what they did to make it unique. This is how you will learn and master 3D. Beyond that, study the art fundamentals: color theory, anatomy, and design. They are the building blocks of your career as an artist, and those will only ever be learned through studying and good hard work!

I hope this tutorial was helpful, I am insanely grateful to be featured here on 80 Level, where I've learned so much myself. Feel free to email me with any questions if you found this tutorial helpful or are confused at all. I'm actively looking for opportunities if anyone needs a helping hand. Thanks all, and best of luck creating awesome stylized stuff!

Cherylynn Lima, Senior Character Artist

Interview conducted by Gloria Levine