logo80lv
Articlesclick_arrow
Talentsclick_arrow
Events
Workshops
Aboutclick_arrow
profile_login
Log in
0
Save
Copy Link
Share

Sculpting Edgy Female Rock Star Character Bust In 3D

Arnaud Jézéquel joined us to discuss the Rebel Rockstar project, breaking down his workflow and sharing insights into sculpting and texturing a semi-realistic, painterly character.

Introduction

Hi everyone! I'm Arnaud Jézéquel, a 3D Character Artist who mainly focuses on stylized projects and hand-painted textures. I started 3D back in 2020 by learning a bit of Blender in college, then I did an art school that taught me the basics of Maya, mainly. 

At first, I was a rigger/animator for a VR video game project at Delusion and for an AR dance experience by Blanca Li. And on the side, I focused on learning character modeling and sculpting on my own.

1 of 2

I've always loved making characters, even before learning 3D. Since high school, I've created posters with characters that I drew for local associations, as I was on the side as a freelance graphic designer then, during my studies.

Now it's been almost a year since I've been in the industry for making characters. I worked at Brunch Studio for the cinematic Why We Fight Back as a Character Modeler, then at Massive Assembly as a Character Artist on the Valorant 30th Agent: Miks cinematic reveal.

I've also worked at Digital Banana this year as a prop modeler on the Clash of Clans Dragon Duke cinematic and other projects as a Character Artist, and I've been teaching character modeling and texturing at ArtFX.

1 of 2

The Rebel Rockstar Project

It began while I was teaching at ArtFX at the end of March. In the course I was giving, the students were to choose between the various illustrations of Yang J's Survival theme character design. And I clicked with the rebel's design.

Yang J

Compared to some other personal works I've done in the past, this time I did not try to match the shading part to the original artwork. I was given the instruction not to make the students paint the lights on the character and give them a more semi-realistic look with more gradients for the hand-painted textures.

As for the setup to work with the reference, I use the Spotlight on ZBrush associated with the movie timeline to quickly get back to the reference position.

Sculpting

As it was for a course in character modeling and texturing, I began in ZBrush with a sphere to show the students how I approach making the face features. I try to blockout some parts: a main sphere for the head, two for the ears, a cylinder for the neck, and a deformed sphere for the upper body part.

Then it's pretty much the basic tools of ZBrush to extrude parts and bits: mainly the move tool for the global shapes and the Clay Buildup brush to add volume. Once I see that the quads tend to deform too much, I DynaMesh for the first time (around the time I got the first shapes for the nose), but I always try to have as few quads/points as possible since at this time, I still focus on the primary/secondary shapes. 

When the global shapes are there, I focus a bit more on the secondary shapes, using tools such as TrimDynamic and hPolish to make flat and sharp parts, and also the DamStandard, mainly on reverse, to get some sharp edges.

Once the main body parts are there, I merge the whole using DynaMesh with a somewhat high resolution to keep most of the details. Then I polish the transitions to make them smoother.

For the clothes, I mask parts of the body that roughly match their shapes, which I extract. Since extracting by default gives polygroups to the upper, lower, and side parts of the extracted object, I delete everything but the upper part to get only a plane that follows the shape of the body.

I then mask some parts (like the sleeves for the vest) and use the group masked function to create polygroups that will be used to create a better topology. So I use the ZRemesher multiple times, keeping groups and in half, to lower the points count while having a nicer flow. Once I'm done with this part, having a low topology plane that still matches the global shapes of the body, I can use the ZModeler to add volume and shapes to the clothes. I also work a lot with the dynamic subdivision on these steps to get a better look at how a more subdivided object would look.

With the ZModeler, I mainly use the extrude tool, insert edgeloops, bevel tools, and the crease tool, to keep some sharp edges even when subdivided. Then I move some points to get back the shapes based on the original illustration. I use this setup of mask, extract, ZRemesh, and ZModeler for most of the clothes and props I make for this kind of personal project. It allows a good matching look on the objects made while them being somewhat optimized.

Hair

For the hair, I always try to blockout the main shapes by sculpting it. Then I mostly use a free curve brush that I found online: DE hair tube.

And then it's placing the main hair strands, following the shapes of the blockout and the original illustration. I tend to use some of the methods I learned from Chloe Worthy's stylized fur tutorial to get more strands easily by masking everything except the hair I want to duplicate. Then I go to the edit mode with the gizmo and duplicate the strands from here. After that, I polygroup automatically so that each hair strand is separate in the same subtool, to control them more easily. And since ZBrush can be a bit overwhelming with all its menus, I created my own toolbar so that I have my most-used brushes and actions easily accessible.

Then I repeat this process for all of the hair, getting from the main hair shapes to the secondary ones and the hair details that follow the volumes.

Retopology

For the retopology part, I use Maya. I get everything I've made from ZBrush, and for this project, I've only re-made the head. The other parts were already optimized on ZBrush. The time I've taken by using the ZModeler, an already optimized curve brush for the hair, and not using DynaMesh or heavy subdivisions, makes me dodge these parts on the optimization side.

Then, for the head retopology, it's the usual tools: put the high-poly head in live surface and then use the Quad Draw to create polygons that follow the head shapes, while having the main loops at the right place.

After that, I get a first pass on the topology. I try to rearrange some parts to get a better feeling and a cleaner "grid" on the topology flow.

I also make some of the meshes that are on the head directly on Maya: the eyelids, the eyebrows, and, for this one, the bandage on the nose and the collar, so that they tend to follow more accurately the polygon flow of the head. 

And then it's more of an organisational setup: tidying up the scene with clean naming before getting back to ZBrush for some adjustments like doing the asymmetric part and projecting some sculpted details (like the folds on the jacket that I did not retopologized) so that when baking, the high-poly version is on the same mesh as the low-poly one. So now this file is the final one that won't move before getting to the textures.

Then, for the UVs, I unfold all the parts by following some sharp edges or hidden ones. And since everything is hand-painted for the end result, I just automatically unfold the meshes because I'll mostly paint by projection on ZBrush, then Substance 3D Painter. There are only a few hair cards and an eye shadow that I've straightened the UVs so that it's more practical, since these ones will be UV-painted. 

I place the various parts on UV tiles to get a UDIM setup. Organic parts on the first line (head, bust, eyelids and eyebrows, then eyes), hair for the second line (fringe, main hair strands, secondary hair strands, and hair cards), and clothes or props for the third line (shirt, shoulder armor, and jacket).

Texturing

For that part, I tend to have some kind of weird setup. I really like to paint directly on ZBrush for a first pass. Because the tools are limited, it's kind of only painting without getting many layers. That, plus the setup I made at the beginning to quickly get back to the view based on the illustration angle, makes it great to place the main information and values where they should be. For the brush, I try to use a square alpha while painting, with a negative Focal Shift to get an even sharper brush stroke.

The way I work is by doing a gradient with a color, then picking a color in between to add a more painted side to the gradient.

But compared to some of my other works, I switched a lot quicker to Substance 3D Painter since I did not completely follow the art style of the original illustration to tend to a more semi-realistic look, and because there are a lot more editing and adjusting tools.

First of all, I bake the high-poly information from ZBrush (that contains the vertex painting and some of the details) onto the low-poly version, and on the ID maps, I select the vertex color option so that I get the painting already made before (sorry if my Substance 3D is in French, by the way, it was by default when installed)

Then I set up a sharp brush like the one I used in ZBrush (a squarish one) to also get some sharp angles. I also work a lot with the wireframe activated with low opacity to see where I could put some light and shadow information.

The main trick I try to get when painting these types of characters, without baking the light, is to have lights and shadows information, but not too much so that you can apply lights for the rendering setup without the model looking off. For example, I did not paint the shadows from the armored shoulder, nor the strap on the jacket, nor the jacket shadow on the skin.

It's kind of a tricky part to get the right amount of shadows and light information. Also, I paint a fake Subsurface Scattering at some places of the skin: the nose, the ears, and the transitions between shadows and lights (the sharp line on the cheek, for example).

For most of the stylization, I tried to put emphasis on the edges. It's mostly visible for the metallic part, but whenever there are some hard edges, I try to have a hard light and a hard shadow. And I put this method kind of everywhere, but to some degree. For the skin part, it's far more subtle, but for the jacket, it's kind of even more present than the armored shoulder, since it's supposed to be worn leather.

I also use what I call the "H" method to stylize some gradients for a more rounded object. You can see it kind of everywhere. I try to pick a color in between two areas and make a sort of H form to merge them more organically while being stylized.

For the hair, since they are a bit curvy in their global flow, I alternated between lighter and darker areas, with a soft gradient between dark and light but sharp ones for light to dark, as I've seen it tend to do that by looking at references, either realistic or stylized. I've also added tiny hard highlights to get a chiseled look.

For the eyes its kind of an accumulation of the various methods I've listed before: the sharp lights/shadows when there's a hard edge, the H method. I've used a white plane to mimic the eye specular, giving it a more stylized look. Also, I tend to darken the upper side of the eye to give it more contrast and a dug effect.

Then I added some layers of adjustments to better mix the various parts of the characters with layer masks in Soft light or Multiply mode.

Lighting & Rendering

For the last step, I go to Blender and render the whole thing with Cycles. I plug the base color maps, the normal maps, and the opacity maps created from Substance 3D Painter into a basic material using mainly the Toon Shader to give a more 2D feel and get most of the information from the Base Color.

Then I split the material based on what they are supposed to be, and tweak the nodes. For the face, I put an extra Mix, adding a Subsurface Scattering setup that is activated with a mask I painted on Substance 3D Painter.

For the metallic part of the armored shoulder, I created a metallic map using the base color and tweaking it in Photoshop, and created an extra normal map using the base color and converting it using Normal Map Online, a free normal map converter.

Then I used a kind of similar type of three-point light setup with three main lights: the key, which I put on the front and is soft with a high degree spread, a blue fill light, and an orange back light that is sharp to better accentuate the outlines. I then added a global top light to have a nice white outline, and an adjustment light to better see the bottom part of the bust.

Finally, I exported the whole with three layers of pass: the combined one, the base color, and the ambient occlusion. I did the compositing of the stills in Photoshop and the turn in After Effects. The result is mostly the basic combined pass, but I put a bit of contrast with the Ambient Occlusion in multiply and the base color on top at like 3% opacity. And the effect on the background (the white grunge type lines) is painted in Photoshop.

Conclusion

I began working on it at the end of March while teaching, so it took about two weeks stretched over a month. For this project, it was really nice working on the optimized version of the face on Maya, trying to get it working so it could be animated, while keeping all the sculpted, stylized details in the topology.

Then the painting process was really fun since it was the first time for a personal project that I went kind of off track with the painting process. Trying to analyze various works of incredible artists to get how they shade a semi-realistic character, and putting it together, I really had a blast doing that. The hardest part was the hair, trying to get a look that is between realistic and over-stylized. It was definitely the longest part to work on.

I can greatly recommend, for the sculpting/modeling part, checking on the FlippedNormals channel, SpeedCharLive, and Chloe Worthy. I've learned so much looking at their content, it's incredible that it'ss available for free on YouTube. Also, I use a lot of the resources of Anatomy4Sculptors, they've got a great free library of 3D assets and other explanatory content. As for the painting part, it's mostly stuff about traditional or digital drawing. So creators like SamDoesArt, Marc Brunet, Pikat, or Une Petite Marie, for those who understand French.

And most of all, trying to analyse and understand various works here and there. Some sort of reverse engineering, it's great to train your eye, trying to see why this part is shaded like this, why there is light there. It's always good to take all kinds of references and immerse yourself in the work of incredible artists!

Arnaud Jézéquel, 3D Character Artist

Interview conducted by Amber Rutherford

Built for Creators. Read by the Best
Partner with 80 Level

Comments

0

arrow
Type your comment here
Leave Comment
Built for Creators. Read by the Best
Partner with 80 Level

We need your consent

We use cookies on this website to make your browsing experience better. By using the site you agree to our use of cookies.Learn more