Obviously Humble Bio
If you’re reading this, you’ve obviously heard of me before, or at least of my work… Back when you read the title of this blog post, most likely. My name is Francisco Azevedo. I’m a lead 3D artist working for Marmalade Game Studios Portugal.
I got into 3D in University, at the same time I discovered 3D art was a thing. Love at first sight, you might say. I have been working in the area for just over 7 years now. It’s something I constantly enjoy learning more about (and also teaching, on occasion). It’s an artform with a lot of depth (haha, because it’s 3D, get it?) and so many ways to do a thing you can never get bored. Even artists working in the same software can have wildly different work methods. And there are also a lot of things to be done in this area: block outs, characters, hard surface modelling, texturing, environments. The list goes on. Working in a small studio, I get to do a bit of each, which helps to keep my work motivating.
My main tools are Blender and Zbrush. Blender is the workhorse for modelling, animation, UVs, etc. Zbrush is more of a hobby tool and a backup for occasional sculpting and blocking out. I do a bit of everything at work, where I usually start from concepts from the 2D artists. The working relationship with our lead artist is so good we even have what I like to call “peer-to-peer concepting”, which is where he can make a couple of concepts from which I extrapolate other assets, with nothing more than some chat about details.
But my main draw is to make characters. I have a big interest in anatomy as well as in figuring out a character’s backstory. The latter is a bigger deal when I’m making something from scratch. There are two ways I go about making character pieces. I either start sculpting in Zbrush and go with the flow, figuring out a theme and mood as I go, or I come up with a simple but interesting concept, then do some sketches to figure out the direction I’m taking that concept in.
This character was made for the latest ArtStation Challenge, the theme of which was Feudal Japan. It’s the third time I have entered this challenge, and the second time I entered the Real-time character category.
In my last two entries I went after a more realistic style, as I relished the chance to develop my anatomy skills. But not being a concept artist myself, I found that I struggled a bit in making a compelling and attractive artwork as a result. Although I promised myself last time I would pick a concept from a better artist as a starting point for the next challenge, I decided to work on an idea of my own devising… again. The result is not as good, but I have more fun in figuring out an original character than simply figuring out the mostly technical aspects of turning 2D into 3D. Only this time, the character I was going to make had to be cartoony.
I believe you need to learn how to do something well, before you do something badly. Or, in other words, learn to follow the rules before you try to break them. Anatomy is an exceptionally hard subject to master. But I felt it was time to start playing with what I had learned and make a more stylized character. The need to make an anatomically correct character became secondary, so I decided to focus more on the story aspect.
A friend of mine from work came up with the idea of making a fisherman’s boy as a samurai wannabe, so I started to think along the same lines. The basic notion was to make a layered character of sorts. So, banking on the issue of how women are still undervalued and underestimated, I decided to make a strong little girl. The other thing that inspired me was the traditional Oni mask used in Japanese theatre. Dressing up as a japanese ogre, or Oni, my character would spook travellers into giving her their belongings. The idea is quite simple, but it leaves room to add details and expand the backstory.
The cat came into picture not because it’s a sure-fire way to gain internet appreciation (it is, but I’m not that clever), but because of a picture my wife found while looking up references. It was a hunk of a samurai with a cute kitten popping out of his kimono. I thought this was too cute and funny not to have. It reminded me of the flying squirrel that travels with Fuu in Samurai Champloo. It was only because my concepts tend to keep evolving, and thanks to feedback from friends, that I ended up using it as part of the Oni disguise and not just a companion animal. Its contribution to the disguise got to be the spooky glowing eyes.
Having settled on the idea, I started the geometry blockout in ZBrush. The girl was the first to be sculpted. I don’t think I quite have a style down when it comes to stylized characters, as I don’t do them often, but her body and head were sculpted pretty quickly. Having done the sketch helped a lot, since she already existed in my head.
Despite an easy start, I was not very confident in the disguised version of the character. I was attached to the idea of a disguise, sure, but the execution felt daunting. Several challenges stood in my way. Japanese clothes feel pretty hard to model, for one (and lo, I already used up the Marvelous Designer trial). The animal skin skirt and the cat both required me to either pick a way to sculpt body hair or create a bunch of hair cards over a large surface. And finally, while I had thought out the silhouette for the girl, the final character was not supposed to be just the girl. I was even advised, after posting the above concept, to focus on the disguised version of the character. So I would basically have to hide what I felt was the strongest part of the character. The whole project started to feel a little shaky at that point.
This stage in projects is where I find myself facing a ton of tasks. Like, “how can I make this look good?”; “that needs retopology”; “that needs UVs”; “that still needs blocking out, then retopo, then UVs”; “what will the pose be? Pose now or later?”. Anyway, to avoid stalling, when one road feels blocked, I pick another one. So I moved to low-poly as that would give me plenty of time to figure out some of stuff still bothering me.
The retopology process is a very important step for 3D artists to practice. It makes a world of difference for rigging, polygon count, avoiding triangles for high-poly sculpting, etc. I did spend some time on this as it was important to get right. The important objects to convert to low poly were the girl, the cat and the mask. Everything else would be modelled from scratch in Blender.
Retopology can be done entirely in Blender or partially done in Zbrush and finished in Blender. The mask was remodeled with extrusions and snap-to-surface. The cat was started with these same techniques to lay down the main shapes, but since the high-poly blockout was a bit… ugly, I ended up finishing the details with simple modelling. The girl’s body was more complicated. I used to draw ZRemesher Guide curves all over the sculpt to direct the topology in the way I wanted. But there are so many things to fix from that method that I jumped onto a new tutorial I found by Danny Mac on Youtube (although he references this other tutorial). The idea here involves turning on the Sculptris Pro mode, filling the object with white and painting in black where you want polygroups to be cut. Then use the Polygroupit plug-in to separate the mesh into polygroups and only then can you predictably use ZRemesher.
An observant reader might notice she has no hands or feet. I had sculpted some, but dynamesh tends to blur together shapes like fingers, making ZRemeshing those parts a bit of a chore. Rather than modelling them from scratch I used the Body Parts IMM brush. I took these hands and feet into Blender and cleaned them up, and joined them to the body. If you think this was a lazy solution…you would be right. It is. But it’s efficient. And getting the pre-made meshes and the new meshes to have the same level of detail and line up is its own task.
It’s important to revisit your model at every step and figure out if you deviated from the original goal. The same goes for keeping an art style in game development. Sometimes the deviation is for the better, and sometimes you lose things along the way. Think of the keywords that ran through your head at the start and consider if they still apply. For me, this meant that I looked at the mask post-retopology and decided it looked too serious. Why so serious, indeed. This is supposed to be a cartoon character, so there’s really no reason to keep such a big part of the model so plain for the sake of authenticity. It should be funny. So I made the mask wonky.
The rest of the model evolved through iteration. The kimono changed patterns and got more tattered along the way. The skirt became a hakama, as I felt it would look better with the rest of the outfit. The cat went from black to the white spotted cat so common in Japan. The pose went from arms lowered to raised to wide open. The spooky tree branches/arms became less spooky and more flowery.
Finally, there was the matter of real-time visualization.
I had never used Sketchfab before. Not quite knowing what to expect from what I always thought of as a free 3D viewer, I signed up and imported my first model (the partly retopologized girl’s body). I was very pleasantly surprised at the fact I could work with PBR materials. This was exactly what I needed! That first import test was a huge relief for me. The eye material even has a teensy bit of refraction, which I hadn’t thought of until I found it as an option in the scene editor. I don’t imagine needing any alternatives for my future work., honestly.
Later into the project, I had a model that was closer to what I was aiming for, but I wasn’t satisfied with the previews in the basic Blender viewer. Going into Blender Cycles was not an attractive idea because it would take ages to set up and to wait for each render. That’s when I turned to Sketchfab on a more regular basis. It was a lifesaver for work-in-progress shots. It helped me figure out what was missing and even helped form the final character, with regards to how the scene was lit.
Only after finalizing my submission to the Artstation Challenge did I realize I had overlooked a huge part of it! I still wanted to show off the girl and cat’s faces, but had no renders whatsoever of them in a 3D viewer. Fortunately, creating a new scene is incredibly easy with Sketchfab. I don’t even mean to shamelessly advertise this platform, but, goodness, did it save my skin. To be honest, I wasn’t expecting to win prizes with this Challenge entry. It was more of a creative outlet and I cut too many corners along the way to truly compete, but I would never forgive myself if I had put out an incomplete entry. So, in less than half an hour, I unfinalized the entry, moved the Oni mask out of the way in Blender, uploaded the new model and created a scene that would show the character in all its glory.
While I laid the post-processing effects onto the main scene (the one with the mask) pretty heavily, this last one was more subtle. SSAO, a little bit of Bloom and a smidge of Tone Mapping to make colours pop. The PBR materials were a cinch to use. Just the right amount of options to give you all the control you need, but not so many as to overwhelm you in menus. I only needed a handful of materials. The ground and lamp each had their own, as they had unique textures, with Height, Albedo, Normal and Roughness textures. The character was simpler as I had already made a 4k atlas for her, including cat, clothes, props, etc. I only made 2 materials for her, to have some parts of the geometry be two-sided (like the hakama/skirt) or one-sided (like the kimono), as needed. Also gone was the dramatic lighting I created with the multi-light + HDR setup for the main scene, and in its stead I used a soft but strong HDR (Muir Wood) to light the character and took advantage of the individual lights to reinforce the effect. I even had one left for the last-minute lamp I added to the base.
At the end of the day, I didn’t win the prize, but I’m really glad I got to make this character and was able to make it look its best in full 3D.