Morrigan: Sculpting, Hair and Skin Production Tips

Introduction

My name is Natalia Stolarz. I was born and raised in Cracow, Poland, but currently, I live in central France.
I have a lot of work experience, but none of it is even remotely close to the 3D world. I studied landscape architecture, had my own company, worked in a ZOO and for IBM. All of them were very, very far from the realms of CGI.

Then, a little more than 2 years ago, I got this amazing opportunity to be able to recycle myself into something new, find myself my ultimate ‘dream job’. At the time, the only thing I knew was that I wanted to become an artist - but how? And what kind? 

By some quirk of fate, I was playing Witcher 3 that winter. So, just like that, I decided that I wanted to become a 3D character artist despite the fact that I didn’t even know this term back then yet. But I really wanted to do it.

Because, well, how hard could it possibly be, right?

Learning path

I started as an absolute self-learner, so basically Google, Gumroad, and Youtube were my first teachers. It took me around a year to understand not only some basics of the 3D art world but also the fact that becoming The Real-Time Character Artist (capital letters intended) might be a teeny tiny bit more of a challenge than I initially thought it’d be. 

With this thought in mind, I started to look for some school or maybe a 3D course, anything that could guide me through this ever-changing abundance of information. And let me tell you that I found something absolutely perfect - Georgian Avasilcutei has just started his mentorship program on Patreon! 

I wasn’t sure if I was prepared enough to be able to work with him but this opportunity was just too good to let it go. Thankfully, I managed somehow to smuggle myself and my fledgling skills under his wings, and he’s been watching over my progress ever since.

First project’s inspiration

I have a big soft spot for the old Bioware games - I was essentially raised on Baldur’s Gate series and then Mass Effect and Dragon Age happened, and I got hooked even more. Remaking Morrigan after a decade from DA: Origins premiere seemed like a very sentimental and tempting choice.
I use Bioware’s concepts for Morrigan though I’m not trying to stay perfectly honest to them. As a practice for anatomy sculpting, I decided to approach creating a likeness of Natalie Portman. No particularly deep thoughts behind this choice, I just like her.

Morrigan is my first project ever, so I’m learning almost all the steps on the way as they come. Let’s just say that I started working on her already in 2019 with very basic knowledge and skills. 

I try not to rush through anything and give myself all the time I need to profoundly understand all the aspects of creating a real-time character, both on the artistic and technical levels. 

As it turned out it has been a full-time job. With lots and lots of over hours.

Sculpting the Anatomy

I started as everyone who has just started this journey should start - with a sphere. 

It was a long and very rocky road to get my sculpt to the point in which she actually started to resemble a human being. I don’t think people realize how much time, energy, focus, and determination (and tears and maniacal laughter outbursts) requires getting through this learning path.
I certainly didn’t.

Before I even started to sculpt, I armed myself with female head scans from the 3DScan store. Having a scanned 3D model opened next to mine helped me immensely to understand the flow of the forms on a human face. In fact, I’d recommend replicating a scanned model or sculpting a likeness to everyone. Trying to achieve some very specific look forces you to rewire your brain to search for the real forms of the original anatomy. Often, we try to sculpt faces in a way we think they look like instead of analyzing our actual references. 

Secondly, I did very thorough research in the fields of human facial anatomy. We have tons of sources we can get our knowledge from if we’re willing to search and devote some time - from cosmetic surgery journals, through traditional sculptors to some amazing industry veterans like Scott Eaton or Frank Tzeng. I get back to my collected materials all the time, and I’m sure I will continue doing that for many years to come. 

In the end, I gathered tons of Natalie Portman photos (seriously, my hard drive looks like evidence in a stalking case). I have even pictures of her life cast mask, which was a great help with understanding the shapes of her face without makeup and without different lighting conditions.
So having my head scans opened in ZBrush, dozens of reference/anatomy pictures opened around, I started to sculpt my first head. 

And then the second one. 

The third. Then the 50th and 100th…and 200th…

It took me around 8 months to get to the level you can see on Artstation. 

Truth to be told, I was also working on the body and clothing when my frustration was getting too overwhelming, and I didn’t know what I was looking at anymore. But I was always getting back and trying to make every next attempt a little bit better - and even with that, I failed many, many times. Until the day I didn’t.

I was very lucky to have an incredibly patient mentor. I cannot count how many times he was heroically trying to explain to me what I was doing wrong and pointing me in a better direction. Annoyingly enough he was always right. (I made quite a puff piece here, didn’t I. Well, good, it’s been well deserved! ) 

Also, the constant support from the wonderful community gathered around him was a huge help. Looking at your WIP through other people’s eyes helps a lot when you’re in the corner and have no idea where to go next.

But still, learning human anatomy is a very long and bumpy process, which cannot happen overnight. There is no magical button here, unfortunately - just tons of references both in 2D and 3D form, loads of research, working from big shapes to small ones, not getting too soon into detailing, watching a silhouette and then sculpting, sculpting, and sculpting. And after that, sculpting some more.

In total, I sculpted 204 heads of different versions of Natalie Portman and her potato-shaped alien evil twin (and these were only the versions I considered worth saving). 

Hair Building

First, I did my research for some references (always!) to have a general idea about what all of the volumes and negative spaces of my hairstyle should look like.

After that, mine three main pillars of the hair creation were:

• Georgian Avasilcutei’s “Real-Time Hair’ tutorial
• George Sladkovsky’s CurveTool plugin for Maya
• Good old manual labor

For creating the hair cards, I chose quite a classical approach - making hair strands with the use of Ornatrix and rendering them to be used as my maps.

For creating the hair cards, I chose quite a classical approach - making hair strands with the use of Ornatrix and rendering them to be used as my maps.

8 different strands were made:

• 1 for covering the scalp - the densest one
• 2 for creating the volumes - mid-density ones
• 2 for variations - low-density ones
• 1 for creating the hairline and the thinnest flyaways
• 1 for the braid
• 1 for the braid’s flyaways

Once I had the albedo, alpha, and depth nicely rendered (Arnold for Maya), I switched to Photoshop to make some adjustments, mostly with the colors but not exclusively. The main braid strand was remade to be tileable, so it could fit seamlessly on the braid’s geometry. Also, the braid flyaways were made to be a bit thicker than the rest of the hair as they needed to be stretched along the braid’s length.

Having it all prepared, I started placing the hair cards by hand. As I mentioned I used George Sladkovsky’s CurveTool Maya plugin, which allowed me to quickly bind a hair card to a curve. It is a crazy handy plugin, with which I could easily duplicate cards, rebuild curves, smooth them, extend, reduce or even create some new cards between the already existing ones.

First, I used the densest cards to cover the scalp, leaving some space for the parting lines. With the same cards, I created my ‘guidelines’ for the general flow of the hairstyle and for the shapes of the biggest volumes (in the end I switched these dense ‘guiding’ cards to the regular volume ones). 

And after that, I just started the process of manually placing the rest of the cards. First, I built the volumes around my guides, then finished the hairline and parting lines with the thinnest strands. Finally, I added the flyaways. I had all of the cards neatly organized and grouped per layers, from the densest to the thinnest ones. 

Once I was roughly happy with the results, I exported the fbx to ZBrush for some touch-ups and adjustments, also at this point, I added the hair rings and placed them on the braid.

I wasn’t limited by any specific polycount budget, but eventually, I ended up optimizing the hair to a little below 100k triangles - including the braid.

I changed my strands and volumes many times until they looked in an acceptable way in Marmoset Toolbag. I was making these changes in Ornatrix, Maya, ZBrush, Toolbag, and Photoshop alike. It didn’t matter if it was about the color, thickness or placing - the whole process involved lots of fixing, paying attention to the references and the details. It involved also lots of getting back and forth and, what’s maybe the most important - listening closely to the trusted feedback until the final effect occurred. 

The eyebrows and eyelashes were also created in Ornatrix. While the eyelashes were left as geometry - the eyebrows were baked on planes. As long as the female model is not a likeness of Cara Delevigne and the eyebrows don’t really change a silhouette of a face - there is no point in keeping them in a geometry form.

Skin Building

The process of creating the look of the skin started already in ZBrush. Once I had my big shapes sculpted, I introduced skin details by using the Killer Workflow from texturing XYZ.
This workflow basically requires assigning the multichannel maps to a plane and wrapping the said plane around the sculpted face using Zwrap - and then projecting the details from the plane to the model. 

I used the details from the red and green channels, putting them each on a separate ZBrush layer. With the help of MorphTarget, I was checking if the main shapes of the sculpted features weren’t distorted so I could, if necessary, morph them back in on a lower subdivision level.

After that, I still needed to clean up a lot of details and add/ replace even more, unique for this particular character. I did it manually just by sculpting them in Zbrush.

The next step was baking all of it in Marmoset Toolbag - basic albedo (from 3D scan store), normals, AO, curvature, and thickness maps. Every one of them was afterward manually cleaned up in Photoshop. 

Having all these pieces nicely cleaned and prepared the magical process of texturing in Substance Painter was just about to begin.

To prevent her complexion from being too flat and boring, I painted some yellows, reds, blues, and greens according to the old ‘three color zones of the face rule. 

It’s also very important to subtly include several color variations - pores, spots, veins, and just some ‘regular irregularities’ every one of us has, paying special attention to the areas around eyes, nose, and ears. Very few of these color layers were purely hand-painted, they are mostly a combination of hand-painting, generators, and grunges.

Once she was all happily colored, I began to work on roughness, specularity, scatter, and translucency. Having my basic maps baked, I could easily use them to emphasize or tune down some features of human skin - which is not by any means flat and smooth. To achieve the desired look, I played a lot with generators and hand-painted layers keeping in mind that all of the elements need to harmoniously work together - pores and wrinkles information should have different characteristics than smooth areas, the T-zone being generally more oily still is not just plainly shiny and all of these areas also need tons of break-ups and variations of their own (viva the grunge maps in Substance!).

Already in Marmoset Toolbag, I added a detail normal map to break the specularity even further.

The final touch was a fuzz. With the help of Ornatrix, I created thousands of small peach fuzz hairs. They were all guided with the surface comb, baked in Toolbag, cleaned up in Photoshop (because who likes women with furry lips…?) and used as the fuzz map.

Those are usually very small and subtle changes, almost unnoticeable at times, but once they are put together, they make all the difference.

While working in Substance, I was checking my progress in Marmoset Toolbag literally all the time. Exporting the maps might be a bit time consuming, but so far that’s the best way I know to fully control the final look of the character.

Presentation

When it comes to the presentation, my approach was quite simple (in theory, at least). 

There are few photographers, whose work I really value and follow. 

One of them is Chris Knight for his wizardry of lights and shadows, and the second is Haris Nukem for his bold approach to colors and vivid storytelling. 

Chris Knight is also the author of a book named “Dramatic Portrait: The Art of Crafting Light and Shadow” - an extremely interesting position to be familiar with. He explains in a very approachable way how the game of lights and shadows is often the ‘make or break’ of the whole picture.

So I picked a couple of portraits made by these two, tried to analyze them and implement the parts I liked into the Marmoset Toolbag scene. 

I also tried to build a bit of Morrigan’s original sassy vibe into the picture but I will work more on this aspect once the whole character is finished and posed.

Once again, it was a process of trying and failing, changing, rearranging, grinding, and fixing.
I don’t even want to admit how many pictures I had rendered before I picked the ones I liked the most. 

Natalia Stolarz, 3D Character Artist 

Interview conducted by Kirill Tokarev