Taj Nabhani did a breakdown of his outstanding character Jjahawa, talked about sculpting, texturing, and presentation, and shared several helpful resources.
Hi! I’m Taj Nabhani, a 3D Artist from Toronto, Ontario, with 5+ years of experience in the VFX industry. I studied illustration at OCAD University, did a post-grad at Sheridan College in Computer Animation and attended CGMA among many other courses, mentorships, and tutorials. I’ve always enjoyed creating a narrative with my work, which led me to VFX.
I was at MR.X in Toronto for 3 years where I worked with some great Character and Creature artists (Atilla Ceylan, Kenneth Doyle, Gabriel Chiang, Nikita Lebedev, Paul Wishart, Gregory Strangis, and Carlos Maciel among many others). I spent my time there working on Vikings, Strain, The Silence, and some other shows here and there. I also got the opportunity to work as Lead on a couple of Nickolodean movies (Legends of the Hidden Temple, and Escape From Mr. Lemoncello’s Library) as well as some unreleased projects.
This project started as part of Bruno Câmara‘s mentorship. I presented him with a bunch of concepts I found appealing. I was looking for something captivating and exciting enough to hold my attention for the entire character pipeline. I also wanted something intimidating and intricate, ideally, something artistically and technically challenging. Mainly, I wanted to use all the skills I had built up in my professional career and make something independently that I could be proud of.
Funnily enough, when Bruno picked the current concept, I thought it would be too easy. I was obviously wrong. There’s a lot in Roman Kupriianov’s piece that you don’t notice at first glance. I think it’s because of the powerful silhouette he uses. When you look at it further, you realize how unique the character’s costume, face, and body language are among other things. It was intimidating to take on something like this, for sure.
Bruno Câmara’s Mentorship
The mentorship with Bruno lasted roughly 5 weeks and the majority of it was spent on how to approach a challenging character and the modeling phase. The piece of advice that really stuck with me was when Bruno asked us early on to choose a world our character could “fit in” (Game of Thrones, Star Wars, God of War, etc.). Having a real-world landmark and not just the general idea of a CG character definitely helped. I tried to go for something inspired by Witcher/Game of Thrones. It was important to have a reference board that reflected this, and PureRef is my favorite program to use for this stage.
I have to say, I don’t want anyone to think they could take just one workshop and achieve a leap in quality as drastic as I seem to have achieved. The truth is my portfolio was more than 3 years old before this piece. During that time I’ve taken multiple workshops with Gio Nakpil, Dylan Ekren, Ehren Bienert, and Pete Sekula to name a few. I rewatched my classes from CGMA with Pete Zoppi and Christian Bull among others, while also watching any Gumroad/tutorials I could find from people like Michael Pavlovich (my favorite source for anything related to ZBrush, and most his stuff is free on his youtube page!), Tom Newbury, and Abdelrahman Kubisi. On top of everything, I kept learning new things at work. It’s been a jam-packed couple years of training, and any creative pursuit is a lifelong journey of failure and improvement.
There’re no tricks in my approach to sculpting, just time-consuming iterative work! It’s all about organization. During this stage of the process, I would only use Photoshop, Maya, and ZBrush. I found that the key to taking on a big project like this is in simplifying every stage as much as possible. Block it out as fast as you can and get a sense of how the forms look. Also, keep in mind that every sculpt is ugly in the beginning, so don’t hesitate to test it out early and often. The book “Anatomy for 3D Artists” from 3dtotal was an incredible help during this phase.
As for sculpting the details of the character, I took that one piece at a time. I would break up the character concept in Photoshop by painting color blocks over every possible different piece of geo and also the pieces I might be able to combine.
The focal length is messed up in ZBrush, so bring it into a program with a real camera and keep checking it frequently. This helps train your eye and gives you a more accurate sense of the proportions. You need to see how the character looks with a 35mm (long shot), 50mm (medium shot) and 85mm (close up) lens to get a real sense of how it will render. You can use whichever Focal Length you want to help sell your character but these are the generic ones I like to use.
Keep in mind, I’m not blocking out every little piece and detail at this stage, only big character-defining chunks. From there, I just Google search all the items to see a more detailed version of them and try my best to recreate each piece. For the parts of the costume that had no real analog, I would just use my design sensibility and a reference vaguely resembling the piece (if I could find anything) while trying to retain the silhouette of the character (in this project, such pieces were the ear ornaments, the crown back piece, and the arrangement of the cape). Also keep in mind that I’m only blocking in “rigid” materials at this point (leather, armor, headwraps). Eventually, I put the character into a symmetrical pose close to her final pose to make life easier when I move on to the clothing stage. It’s a slow process but the devil’s in the detail.
There were multiple times during this project where I almost abandoned the entire sculpt and focused on the face only, so it definitely got a bit more attention than the rest of the piece. It didn’t turn out a detriment to the final product, though. The face of your character is the most important part that you need to get right and working on it was a constant process. Up until the final render, I kept making small tweaks here and there.
The main reference for the face was the actress from The Wolverine movie Rila Fukushima. Early on, the character resembled Rila quite much but in order not to step away from the concept, I ended up abandoning the likeness. However, it’s still important to have a real-world base to start from before you start stylizing the character to match the concept. It grounds any changes you make in reality even if you don’t capture the exact likeness.
The face is complex as it has layers of subtle forms that can change the look drastically when slightly nudged. You need even more than just a competent understanding of the major and minor forms in order to create a decent portrait. The Asaro Head can be an incredible asset to have close by, as you can constantly check your planes against it.
As for the details around the face and lips, I used some Texturing XYZ maps to create a road map for such details quickly. At the time I was working on this stage they had just released a tutorial for their new multichannel face packs. I loved how quickly they helped to get nice results with the surfacing maps but I didn’t have the money to buy a new multichannel pack. So, I used a similar process with a self-made multichannel texture from their old packs. After you get the grain of the skin and surface details, it’s important that you go back to your sculpt and integrate these elements by sculpting a bit on your character.
The hair wasn’t going to be a part of the model originally, but when I projected the textures of a real face onto my sculpt, I got some hair around the temples from the images and liked how it looked, so I went with it. As for the tech behind it, it’s 100% XGen. There’s a really great and free tutorial series from Jesus FC on XGen as well as his Patreon.
As for the eyes, that turned out to be a mini project on itself. Originally, I had a very basic setup for the eyes. After trying out a couple of different arrangements that didn’t feel right, I ended up going with Tom Newbury‘s new setup that can be found here. I made some slight variations to achieve a look I wanted and applied his knowledge to VRay (he uses Arnold). He had some really great tips and techniques on integrating the eye into the face by using a meniscus, caruncle, and noise on the cornea. His entire series is great and extremely helpful.
Clothes in Marvelous Designer
The most important thing to know about Marvelous Designer is that it’s mostly a block-in tool. It simulates clothing really well but it’s still basic and you really have to push the program and your computer to get anything close to the final result right out of MD. You need to bring what you get into ZBrush or Maya and tweak the geo/ sculpt for finer folds and details to get it production-ready.
I’m still learning and improving my Marvelous Designer workflow but the developer’s website has some fantastic resources for learning the program. My personal favorite resource for learning Marvelous Designer is the Gnomon tutorial by Paul Liaw and this tutorial from Seth Nash. If you can recreate this complex piece, you’ll be able to make almost anything.
There’s also an amazing new tool in MD that creates a pattern from UVs – it can speed up the process a bit for basic things like taking measurements and creating a fitting slope. I used this new tool for the star-shaped neck of the cape, I sculpted the shape in ZBrush and brought it into MD as a UV pattern. From there, I added the length and simulated the weight of the points.
When making a character with a lot of layers like Jjahawa, I would simulate one piece at a time. I would then export it as geo and reimport it into MD as a part of the avatar. This approach was less chaotic than the layering system or freezing, as I was never able to get the layering system working right (the more layers you add, the buggier it can get). I also found that clothes can interact better if you use this Morph Targets trick shared by Olivier Couston. Much like the modeling process, it’s all about working on one piece at a time. When everything is ready, bring it all together in ZBrush to see how it looks. A great resource for that can be found here, also from Olivier Couston.
Roman Kupriianov had already established the color palette in his incredible concept, and it was my job to retain as much as possible and stay true to the concept. The first step in the texturing pipeline was to identify my main materials which you can see in my initial block-in with the colors. I use these as a base and build on them using some real-world references.
The most important resource I’ve found for developing a basic understanding of Character Materials and Rendering is Xavier Coelho-Kostolny’s YouTube channel.
For the head, I used the same method as for the displacement but I took my cross-polarized images from Texturing XYZ and applied them to a plane, then wrapped it. This helped me to get around 75% of the face and hands done within a day. See the workflow here.
Once I had all my patches from the wrap ready, I layered them over each other in Mari and masked out all the stretched or unusable parts.
I got a pretty raw but clean base to start with and could begin cleaning it up from there – and that was an extensive process involving cavity maps processed from the displacement and other procedural layers. Tom Newbury and Abdelrahman Kubisi were excellent resources for this stage of the project. They have easy-to-follow tutorials.
Luckily for me, I only had to make the face and could mostly ignore the ears which are a whole new process in itself. Once I’m satisfied with the skin, I can move to the costume.
I haven’t stressed it enough, so: REFERENCE IS KEY for every stage of the process, especially when it comes to the costume of a character.
In the beginning, I spend most of my time in Substance Painter (I prefer it over Mari for inorganic objects) and on Substance Painter YouTube channel because it’s extensive, cohesive, and easy-to-follow. I can quickly block in materials and get the right amount of surface definition for all the pieces while retaining unified aesthetics of the piece. It’s also important to add small details to keep the viewer’s interest and remember that every piece on the character has its own history. This is something I learned from working on Vikings. You need a good amount of wear and tear for any costume to give it a more lived-in, real-world feeling. You also have to take into consideration how differently each material reacts to various conditions (compare cracked leather to sun faded/dirty/stained cotton, for example).
I start very simple by going around the entire object and just applying basic materials that roughly match different pieces of the costume. Here, I try to work procedurally as much as I can and rarely paint anything for the costume. Instead, I use the Mask Generators within Substance Painter to create wear/tear/dirt and only go back into the key areas to give them extra hand-painted detailing.
Substance Source was a vital part of this process. I could find anything I wanted there and test it out quickly. For example, I could test out a variety of leather materials on the boots and the belts without having to leave the program.
For the patterns on the vest and skirt, I mainly used Photoshop. First, I implemented them within Substance Painter to create height information and then created a mask to introduce a difference in the embossed leather giving a carved feeling to the pattern. I also use Photoshop to create an opacity map for the worn threads running along the edge of the cape.
At a certain point, Substance has its limits, especially when you try to take on an entire character within one file. You cannot paint across UDIM patches within Painter. Also, the files can balloon in size fast, especially if you bring in your own textures to use within Painter. In a professional project, I would’ve broken it up into segments and worked on each part individually. Either way, you get to a point where you need Mari so that you can create a unified layer for your textures across UDIMs. This might be as simple as a subtle gradation across the whole character from bottom to top or a color gradient on some parts. Also, in my experience, Mari can handle larger files at higher fidelity than Painter. In any case, they’re pretty even and both have strong pros and cons. If you want the best results, use both.
Rendering & Presentation
I don’t like to worry about polycount when working on personal projects, especially since my end goal for this piece was a static render. In the end, it had roughly 500 000 polys.
The biggest hurdle for me was the rendering process. Originally, I wanted to go nuts with the resolution of this character but ended up optimizing and cutting a lot of details. That was probably for the best as some things were pretty superfluous. I remember the biggest roadblock was getting the project to render on my computer which would either crash or take ages to complete. I was pretty frustrated at that stage. I was initially using Arnold and really enjoyed how fast and easy it was, however, I found it a bit limited and slow for a full character.
Then, I switched to a trial version of V-Ray Next recommended by a colleague and loved how robust the renderer was. I haven’t used V-Ray since roughly 2016 and was excited to test out the new GPU-based side of the engine. I was not disappointed and had tons of fun playing with the shaders within the new superfast IPR.
Once I had plugged in all my materials into a basic shader I could test out my lighting quickly with the IPR. I started with a basic 3-point setup and an HDRI. The goal here was to create an atmosphere with those 3 lights that would match the mood of the character. It’s a versatile and at the same time simple setup if you play with the intensities, distance, and angles of the lights.
Since it was based on a concept, I wanted to get as close to the original pose and camera angle of the illustration as I could. I went for a medium- and close-up at roughly the same angle, just to show off a bit of the detailing. I also played with some other angles but didn’t stray far from the original concept.
There was a lot of learning throughout this project and sometimes I had no idea how to continue. It gave me the opportunity to develop some techniques and learn a lot about the process of creating a cinematic-quality character by scraping through frustrating times with a lot of research. I think it’s clear that there isn’t a single workshop or class that will develop all your skills quickly, and for that, you have to put a lot of effort into work. I outlined maybe only half of the resources I used but the ones I’ve mentioned were the most helpful during my process. I’m looking forward to my next project and will take it further than my last one.