I’m a multidisciplinary artist who grew up in the cold mountains near Seattle Washington surrounded by tantalizingly mysterious forests covered in clouds, rather like the one in the opening sequence of Majora’s Mask. As a kid, I never did find a Deku forest temple buried somewhere in the pacific northwest, not for lack of trying. But Majora’s Mask, and the way a multitude of interwoven characters defined the very shape of the game stuck with me. And filled my childhood sketchbooks.
The first time I ever wrote code was to build a website where I could host my Legend of Zelda fanart. My passion for characters has remained unwavering throughout the course of my career. I graduated with a double degree in computer science and art, with an eye-opening variety of experiences in everything from conceptual art, data visualization, digital interactive installation work, and machine learning, to more traditional drawing and sculpting. But no matter the subject I filled my sketchbooks with “silly little characters” (as my professors termed them).
After graduating I followed my passions and spent over six years on The Alice Team, designing and sculpting quirky magical characters. The Alice Project is a 3D coding environment where students learn how to program in Java by creating games and animations. Alice is an amazing program that allows educators to easily incorporate multidisciplinary projects into their computer science curriculum. Building cool games with only a few blocks of code and art assets was something I could only have dreamed of when I was a kid, and seeing students as young as middle school being able to do that in Alice is a thrill that never gets old. I believe in the power of providing people with knowledge to craft their own stories, and for me, the Alice Project educating students in coding, art, and storytelling has always been a tool to do just that.
New Orleans Victorian Character
The character started as a discussion with my writer friend Reyna Princivil on what types of new characters we would most like to see in games. We brainstormed the concept design, including thoughts on what kind of abilities this hypothetical character could have, and Reyna wrote the backstory. Black women are extremely underrepresented in games, so picking two of the biggest pop culture moments in the last few years: Lemonade and Black Panther, and using them as influences to imagine a game character was very intentional.
My main goal was to create a character who could inhabit the New Orleans Victorian gothic aesthetic of Beyoncé’s Lemonade. And, using the costumes of Black Panther and Ruth E. Carter’s work as inspiration, I looked to art from Nigeria, Kenya, and other African countries to weave in similar cultural influences also present in Lemonade. For the character herself, I chose Lupita Nyong’o to challenge myself to work towards facial likeness. The overall design pays tribute to Beyoncé’s iconic giant Formation hat. Basically, I picked a subject I was extremely passionate about, did a ton of research, and got to work creating an homage to brilliant artists I admire: all in the form of a wish-fulfillment hypothetical video game character with a presence steeped in lore who hopefully looks like she could magically kick ass in a fight. It was the Game Art Institute Bootcamp that taught me how to use current game industry tools and artistic technique to bring this dream character to life.
Learning Character Art at GAI
Most of my previous character work has been in animal and creature design, so while I know a lot about four-legged anatomy and way too much about bird wings, prior to the GAI class I had never sculpted a realistic human face. However, in my personal sketchbooks, I’m always drawing portraits: friends, celebrities, characters, anyone whose face has a lot of personality. My idea of relaxing after a day of sculpting is to sit and draw faces for an hour. I’m really enthusiastic about capturing the likeness. Not just the physical details, but how they feel, their body language and movement. My favorite thing is drawing from gifs. I have over 100 gifs of Nakia from Black Panther saved. For each one, I’d sketch it freehand, then pause it and bring it over to photoshop and trace the image. Then compare the two to try and figure out what lines make Lupita look like Lupita.
Ryan Kingslien’s approach to teaching anatomy was the main reason I chose his GAI Bootcamp. When I started out in digital sculpting, Ryan was the artist we all wanted to grow up to be one day. His art pushed the boundaries of fantastic technical anatomy while maintaining expression and personality. Learning from him through videos and getting live feedback gave me the chance to translate my obsessive 2D drawings into 3D sculpture. I really liked how the focus was on getting the technical details and challenges to work within the whole art piece. Rather than letting technology define art. It’s not just teaching the technical basics of like making an eye or face, it’s how to artistically direct that process. While also still knowing what parts of the face go where and being able to recognize the small subtle bits that determine likeness. Plus it’s a fun party trick to be able to point to a part of the face, and recite from memory “oh yeah, that’s the (insert unintelligible mumble that vaguely sounds like the proper term)”.
From the start, I knew I wanted the silhouette to be broad, and to spread out like the giant Formation hat and Victorian-era hoop skirts. I took cues from Lemonade, and used historical references but modernized them. The design elements on my dress are a mashup of different bits of costumes seen in Lemonade. I wanted the dress to feel improbable, but functional. The type of thing one might find in an over the top action game. The crinoline and skirt split in two down the front, with a short underskirt, to allow for more movement and give the character the ability to kick someone in the face if necessary. There were multiple layers to this dress and I intended for each one to tell a story. Despite having a background in sewing, I think this dress would be impossible to recreate in actual cloth. It would be heavy and impractical, and you couldn’t walk through narrow doors because I made the crinoline wood. But that’s why digital is so much fun!
It was my first time using Marvelous Designer, and one of Ryan’s tips was to think of Marvelous as more of a sculpting tool. I studied Marvelous’ tools and strengths and then used those in creative ways to get the result I wanted. So, to get the wide shape I decided I needed to customize a fake stand-in crinoline that could replicate the look of heavy fabric padded on top of stiff hoops. I sculpted a copy of how I wanted the silk to drape rather than use the original crinoline itself. I took the blocky beetle shaped construct and imported it into Marvelous alongside my character. The underskirt fits the beetle crinoline snuggly, except for the little peplum in front. There is a tiny strip of fabric running around the half circle edge of the rim of the crinoline that I cut off the underskirt and ‘froze’ to hold it in place. This allowed me to keep the underskirt snug even when I deleted the back piece of fabric and replaced it with the wavy triangle folds that you see on the final dress.
The overskirt was stitched on top of the underskirt in eight panels. To get the drape of each panel right I studied a lot of historical patterns from the eighteenth and nineteenth centuries. The bodice of the dress is a pattern taken directly from an 1860’s ball gown. Without the historical reference, and a guide on how to take proper measurements, I would not have known how to get the bodice form fitting enough for it to hold up on its own without shoulder straps or any of the usual corset type inner structure.
Fabric Simulation & Texturing
Marvelous has fabric simulation attributes that are thankfully specific enough I was able to get the physical properties of the cloth to look how I wanted without much adjustment to the simulation. On the other hand, during simulation, everything stretches more than expected for actual cloth, so I had to take that into account. The stretch made things like the bodice especially difficult. I utilized the strengthen and freeze tools during simulation to control which parts were holding the entire dress up and which parts were draping. Unable to structure the bodice with anything other than fabric in Marvelous, I simulated the entire top half first and then froze it. With the bodice held in place, I was able to make the skirt without the heavy fabric pulling the rest of the dress down.
In this way, I left very little of the dress to straight simulation. I picked and pulled at the fabric with the little hand tool and tacks during the simulation to get the folds exactly how I wanted. The entire time I was using reference and being very picky, looking at how folds in photos would actually work with the type of fabric I was using.
To texture the dress, I made my own silk material. I downloaded a few texture photos of dupioni silk and made an alpha for the base weave. Then I added multiple layers of color and texture to simulate the trademark uneven weft. I also added tiny tiny flecks of metal to make it slightly sparkle in the light, because I like sparkly things. Finally to fake the iridescent look I used the curvature map to bake alternating colors into the folds of the fabric, which in real silk would change with the light. For the base roughness and metalness, I actually made it 50% metal since that gave the sheen I wanted, but everything I read online said this was a no-no. And I thought my dress might look like ‘alien fabric’ in the final render until I listened to a lecture by Pauline Boiteux and learned she does the same thing. After that, I felt much better about standing by my weird half-metal material!
Marvelous Designer-ZBrush Pipeline
Marvelous made it possible to do a very complex dress in a relatively short amount of time, and that really impressed me. However, the challenge was to use the program to sculpt my own vision rather than letting the technical constraints shape the design. I did choose a few details in the dress based on my knowledge of Marvelous’ strengths. For example, the ruffles across the top of the bodice, and the drape of the back of the skirt. For my piece, I stuck to creating patterns in Marvelous and simulating them. What you see in the high res version was taken directly from Marvelous, with no editing in ZBrush beyond adding thickness to the cloth and piping to the bodice. I mostly used this process because I wanted to learn as much as I could about Marvelous, and really test my capabilities in the software.
I found a lot of the historical dress patterns difficult to replicate because getting complicated folds proper was almost impossible for me in Marvelous’ simulation. I would like to work further on this by adding ZBrush into the pipeline. I believe it might be possible to first sculpt the basic shape in ZBrush and then bring that into Marvelous for simulation. This would be necessary since the main thing missing in Marvelous is the filler and padding that is stuffed into things like sleeves and atop crinolines. Making the filler shapes in ZBrush first could make the fun bustles and layered poofy silhouettes seen in Victorian dresses possible.
Challenges & Feedback
Maintaining an efficient pipeline while experimenting with software new to me was the most frustrating challenge. Making technical decisions and only realizing after I got to the next step that my previous choices disrupted the workflow and required lengthy fixes meant some parts dragged on more than expected. Some of this I had circumvented by creating three quick test dresses in Marvelous, but if I could do it over again I would have taken one of those dresses all the way to the retopo and texture phase before starting on the much larger, much more complicated final dress.
The texture and normals of the overskirt were especially tricky. I am still unsure if I picked the optimal way to do it. For one, the way the panels overlapped each other with lots of folds made retopo difficult. Next time I would ensure the fabric thickness setting in Marvelous is set to the final thickness I’ll give it in ZBrush. This way when the folds simulate there will still be enough room between them to add thickness and make baking easier. There were too many places where the fabric collided against itself or folded then covered itself. This also happened in the ruffles on the shorts. You can still see on the final model that since I based the retopo off of the 3D overskirt rather than the unfolded 2D model, the UV’s do not show the fabric pattern naturally. It looks more like an overlay across the top of the entire dress than a folded pattern. On the other hand, the polycount is a lot lower than if I had included all the overlap fabric in the low res.
Writing out all the new technical things I learned on this project would result in a very long list! I think the most important thing I took away from GAI was how to observe and breakdown references visually and apply that knowledge to a sculpt. Specifically, things like Ryan’s explanation of anatomy and how practicing and picking out anatomical details on reference trains your eye to see those details missing in your work. And similarly in substance painter where it’s all about layers and training yourself to notice how materials are built up.
In addition to GAI, I’d like to thank sculptor Ethan Gagorik and illustrators Bob Kato and Forrest Card of the Drawing Club, who in the months after the Bootcamp ended critiqued my textures and renders, and helped push me towards finishing the final stages of my character! I had a vision of getting this character into an environment (the background is based off Fort Macomb as seen in Lemonade during the song “Daddy Lessons”!) and they helped me chase that.
Check out this nice Flower Pattern cloth from Quixel!