Realistic Evangelion's Asuka Reimagined in Blender & Substance 3D Painter

Introduction

Hi, my name is Wahyu Nugraha, I’m a 3D Artist from Indonesia currently based in Malaysia. I’ve been involved in the game industry since 2011, after graduating with a Multimedia major from college. I decided to change my path after finding an interview article by Sze Jones on the Pixologic website regarding her involvement in the 3D game industry and knew straight away that the game industry was what I was looking for. I’ve been playing video games my entire life but never once thought about it as a full-time job. So I began doing more research and was involved in online forum discussions to teach myself about this new-found goal.

Though at that time, my character knowledge was not able to land me a job, so I tried my luck at environment art and luckily enough, I got my first job as an environment artist, all the way until now. During all this time, character art has been living at the back of my mind, being my first love in the game industry, but I treat it more as a hobby that I visit once in a while during my free time.

Hard-Surface to Character

Asuka started as another playful project for my character art hobby, providing a change of pace from my usual hard surface work at my day job. I was inspired by Shirogane-sama's portrayal of the character, which looks exactly like what I'd imagined if Asuka's character were to exist in real life.

What inspired me the most to shift this project into high gear was Jinx's portrait by Joshua Gouw, simply because it looks too amazing not to want one too! So, I gathered many more references and compiled them into PureRef, from Shirogane-sama's cosplay photos to other artists' renditions of Asuka, in order to gain different ideas and approaches to create the same character and its art direction.

Head

For this project, I'm not aiming to achieve a perfect 100% likeness of Shirogane-sama. What's important to me is capturing Asuka's characteristic attitude; her sassy, proud, and strong presence, as well as her past trauma and freckles!

I began with a simple remeshed sphere for the main head volume and two flattened spheres for the ears. Then, I sculpted the facial features by overlaying a photo reference on top. PureRef's transparency mode really made things faster here, without requiring me to set up a camera projection in Blender.

Skin

Once I was happy with the face proportion, I proceeded to build a proper topology for the face that I could UV unwrap and map to Texturing XYZ multi-channel face. For this specific project, I used Female 30s Multichannel Face #07. There was no micro skin detail sculpting involved, ensuring that the color and displacement matched 100%.

The primary purpose of this UV channel is to project the Texturing XYZ image data onto the mesh. Then, I created a second UV channel within the 0-1 UV space for the final texture to be applied onto, and transferred the image data from UV1 to UV2, including all three maps that came from Texturing XYZ: Albedo, Utility, and Displacement maps. So the only texture baking I did for the head was to transfer the maps, with no high-poly to low-poly baking.

I performed all the texture clean-up in Substance 3D Painter and added further details to make my Asuka unique, based on Shirogane-sama's cosplay makeup. I also decided to replace the plane geo eyelashes with curve eyelashes to add more realism to the character's face.

At this stage, I added a standard environment and light setup in Blender so I could create and test the material node system for the head. There was a lot of tweaking happening during this stage, especially for the Utility map's red channel (hemoglobin) and green channel (melanin), with constant back-and-forth adjustments between Substance 3D Painter and Blender's material node to achieve the result that I was satisfied with.

The material node itself was not overly complex. Texturing XYZ provided a handy guide on how to use the Utility map with Photoshop, and all I did was replicate the process using Blender's material node system.

Eyes

For the eyes, I decided to use Auto Eye from BlenderMarket since it has all the features I need and more. It significantly saves a lot of time for my next character project, as its reusability is almost endless!

I created a small shader for the tear line mesh to fade off the edges and add a slight noise bump to the surface.

Hair

The part that took me the longest to tweak, probably even more than the skin shader, was the hair! The primary tool I used in creating the hair was the Hair Tool. I won't go into details here, as the add-on author has a comprehensive wiki that teaches all about it in great detail.

First things first, always start with the blockout mesh to get the overall feel/flow of the hair. Then, I begin layering the hair cards from the bottom (the densest hair per alpha card) to the top (a few strands per alpha card).

The Plugsuit

Mesh

Much like the head, the body also started with stretched and squeezed spheres to build the main volume and determine the major proportions of the body. I then remeshed them all together and continued with a more refined, yet loose sculpt, all within Blender. The purpose of this mesh was solely to extract the mask from it.

The workflow for the plugsuit is all about Mask > Extract > Quad Remesher > Solidify. I repeated this process until all suit parts were created, and used sculpting or traditional modeling for additional details.

Once I have all parts created and details added, the next step is to create the low-poly mesh that will eventually be textured and rigged. For this task, I use RetopoFlow to help streamline the process. Unfortunately, there's no shortcut to getting a good low-poly mesh other than being patient.

For Asuka in particular, I split the suit into three material IDs: Body, Hands, and Legs. This was because, earlier in the project, I had planned for swappable bare hands and legs (in addition to the full-body plugsuit), similar to one of Shirogane-sama's cosplay costumes. However, I decided to scrap the idea, but the three-ID planning remained until the end.

Texture

A good practice for texturing is to build the material layer by layer; don't add dirt, scratches, and other surface imperfections before achieving a good-looking base material. Refer to the reference images often during the texturing process, because relying solely on memory may cause you to miss subtle details, hue shifts, and micro-roughness variations that add more visual interest to the final texture.

Always check individual channels, especially Albedo and Roughness, throughout the texturing process. The more interesting they are when viewed individually, the better they will be when combined with other channels.

The Spear of Longinus

As primarily a hard-surface artist, modeling the spear was relatively easy, taking me approximately 2.5-3 hours to finish, from blockout to final texture. I prefer to keep my mesh non-destructive as long as possible by utilizing Blender's modifier stack system. This approach not only makes it easier to make changes parametrically but also ensures consistent measurements. For hard surface tasks, my go-to add-ons are ND, Kekit, and MESHMachine. I won't explain them here, as they have very informative wikis available, so check them out if you're interested!

Here's a GIF that shows the steps I used to model the main shaft non-destructively: 

Once the main shaft was completed, I duplicated the mesh and applied the modeling modifier to create the head part of the spear, using traditional modeling methods.

As the spear has the same material as Asuka's red suit, I was able to export and reuse the smart material to apply it to the spear with only minor tweaks needed.

Rigging And Posing

Rig

With all meshes completed, it's time to do rigging. For this task, I used Auto-Rig Pro and Voxel Heat Diffuse Skinning. Of all the workflows mentioned here so far, rigging is probably the most alien to me. But luckily, they also have a very detailed wiki to explain anything you need to know regarding the toolset.

However, there's no magic button to get an error-free rig for my character, so I had to learn and explore how Shape Keys work in Blender. Essentially, it's a tool that lets artists correct a mesh's deformation in certain poses that can't be corrected using weight painting alone. For the head, I ended up adding Shape Keys for the eyes and lips, and for the body, I added Shape Keys for the fingers and pelvis area.

Pose

I picked 10 poses that I felt best captured Asuka's character and recreated them for my final shots. It was both fun and challenging. Fun because it felt like playing with an action figure, but challenging because the rig I made for Asuka wasn't error-free even after all the tweaks, and it had limitations due to my limited knowledge on the subject, but it still did a pretty good job for all the poses I planned on recreating.

Here are a couple of fun behind-the-scenes moments from the posing session!

Light, Camera, Render!

The last step, and arguably the most important one, is to light the scene, set the camera composition, and hit "Render". For this project I used Blender's internal renderer, Cycles, to do the final render.

One thing I'd like to share here is how I organize my scene, and I found that this is the best way for me. I have a separate scene for my modeling, rigging, and rendering.

Modeling Scene: contain all the blockout, high poly, low poly, and all the utility meshes I used for modeling.

Rigging Scene: contain only the final low poly and the rig.

Rendering Scene: contain a linked low-poly mesh and the rig so I can update the mesh and the rig on the rigging scene, and the changes will be replicated in all instance of them on my rendering scene.

Lighting

Most, if not all, of my scenes start with a 3-point light setup: key, fill, and rim. For those who are not familiar with photography, I can describe these in simple terms:

• Key light: the main source of light to illuminate the center of attraction, most of the time it will be the head region.
• Fill light: a light to fill in the deep shadows cast by the key light, to bring in detail to otherwise dark areas.
• Rim light: the purpose of this light is to draw an "outline" of the subject by highlighting it from behind, creating more separation between the character and the background. I tend to set this light with high intensity.

But consider this a general rule, because besides these three lights, I also use HDRI to illuminate the scene globally, a face light – a small light to highlight the face's spec from a certain angle, a hair light – similar to the rim light but focused on the hair to give Asuka's golden/red hair SSS more glow, and I also choose to use two rim lights with different intensities to illuminate the left and right outlines of Asuka.

When setting up lighting for my scene, I like to think of it as painting the subject with lights. Gradually, I tweak lights in the scene one by one, constantly asking myself which parts of the model haven't been "painted" yet (due to the lack of light to shine on them), and add or adjust the corresponding light accordingly. At the same time, I try not to add too many lights, which can result in the render looking flat. Shadows are also important for bringing a sense of 3D depth to the final image. I play around with light and dark areas to complement each other without any one area standing out too much.

Camera

For the camera, besides the focal length setting and the render resolution, which is a 1:1 ratio (square format), I use the default camera setting. For my renders, I try to make each shot unique from one another and feel dynamic by tilting the camera left or right, using low or high angles, and adjusting the focal length. For this project in particular, the poses and shot compositions are heavily inspired by Shirogane-sama's photo shoot.

When setting up the shot, it's great if your software of choice can display multiple viewports, and in my case, Blender provides this with ease. This is how I usually set up my workspace to facilitate my needs while tweaking light positions and camera compositions; the bottom left viewport is handy for checking the final image preview.

Rendering

With all meshes completed, light setups checked, and cameras ready, it's render time! I'm using Blender's Scene feature and creating each pose in its own scene. The benefit of doing this is that it keeps everything tidy, and all I have in the current scene I'm working on are the elements related to my current shot/composition, without having to create multiple files.

3..2..1..Render!

I added a Glare effect from Blender's compositor, then exported the image and brought it to Affinity Photo for the final polish. By using Tone Mapping Persona, I created a custom preset that I can reuse for all my rendered images, speeding up the process.

The adjusted properties here won't work exactly the same for each of my images, just to give a rough idea of what's adjusted: I bumped up the local contrast, brightness, and saturation.

Conclusion

And we're done! I started this project early in 2022 for fun during my free time on and off, and who would've thought it would end up being my most complex personal project to date? I learned a lot about organizing a character's scene, creating hair, rigging, and final composition with additional post work. The main challenges, I would say, are the technical aspects of character art. I almost gave up halfway through, but I'm glad I still managed to push it until the end.

A final word to everyone who managed to read all the way to the end: art is a very dynamic thing and there is no right or wrong way to do certain tasks. What I've shown here is just one of the many approaches one can take to complete such a task. I certainly wouldn't have known if this method would work or not if I hadn't taken the first step in doing it. So my advice to anyone with ideas or dreams of creating character art is to take that first step, and your experience will show you which method works best for you personally and help you continue to grow from there.

Thank you for all the positive feedback and to 80 Level for the opportunity to share my making of the Asuka project. Thank you for reading this article to the end, and I hope you find it informative in one way or another.

Wahyu Nugraha, 3D Artist

Interview conducted by Theodore McKenzie