VUE without competition
Can you please give us a walkthrough how to implement this into Maya? would be super helpful. Thanks a lot.
Jaco Herbst who worked on such titles as Dota 2, Counter-Strike, the Batman series and the Total War: Warhammer series shared his experience in creating awesome characters for video games, tips, tricks, and preferences.
Firstly, I just want to say thanks for giving me the chance to share some thoughts, I appreciate the opportunity.
My name is Jaco Herbst. I am a self-taught freelance character artist, living in South Africa. I’ve been working mostly in the video games industry for the past 10 years, during which I’ve had a part in games like Dota 2, Counter-Strike, the Batman series and most recently the Total War: Warhammer series.
I got my start in the industry at a fairly young age, starting work pretty much straight after high-school. I have an online community, Polycount.com, to thank for that. I discovered the forum as I was trying to learn about modding. It was mostly dedicated to creating and sharing custom skins for games like Quake and Half-Life at the time. Discovering what this community was about blew my 15-year-old mind, and pretty soon I started posting my own work and participating in art challenges such as Dominance War, all the while getting excellent feedback and advice from generous industry pros.
The invaluable experience and exposure I got from that community led to me having somewhat of a portfolio by the time I finished high-school. An indie developer saw my work on the forum and got in touch; this was right around the time I was thinking about applying for university. Luckily I never got as far as actually applying, as the freelance was steady enough for me to take the plunge. I was very fortunate that my parents were so supportive of this strange career path!
Video Game Character Production Pipeline
I think there are a few key factors that help keep the style and quality level consistent.
The first is the use of a blockout mesh. This is a pretty simple mesh but it does a good job of matching the concept art when viewed from a distance, capturing the major masses and silhouette. The blockout is also used by animators/riggers to catch any issues with intersections and the like. To create the blockout it’s common to start with an existing basemesh and then building around that, reusing parts from previous characters if time is limited.
After the blockout is approved, work begins on the final detailed sculpt. It’s helpful if there’s an existing sculpt to reference, this helps keep the style and detail level consistent. Usually, there’s some back and forth with revisions at this stage, as the artist tries to capture the look and feel of the concept art. The temptation to detail can be hard to fight here, but especially on a game like Total War: Warhammer it’s important to focus on creating larger areas of contrast. This is done with the material breakup, for example, large areas of exposed skin vs. large areas of the cloth. These large areas of contrast help maintain readability when the characters are viewed from a distance or are seen in large groups.
It’s also helpful to start thinking about how the in-game mesh will be created while working on the sculpt, figuring out where texture seams will be placed, which parts will be mirrored or re-used, where to focus detail, where parts can be combined and simplified etc. Games have fairly strict texture and polygon budgets, and spotting trouble areas early on can save many headaches later. Personally, this is the part of the pipeline I like to spend most of my time own, as everything else is dependent on a well crafted sculpt.
After the sculpt is approved, an in-game low poly mesh is created and unwrapped. I would advise people interested in creating efficient game ready meshes to study some of the content that’s been released by studios, for example, the Dota 2 characters can all be downloaded for study here.
And models from Epic’s Paragon can be downloaded here.
The final step is baking and texturing. I find that with the advent of programs such as Substance Painter texturing has become fairly straightforward, lifting much of the technical burden and allowing artists to focus on the important stuff. Something like Smart Materials and material presets, in general, will do much of the heavy lifting for you, getting 90% of the result with minimum effort. Most of the time I spend on texturing is in getting that final 10%, as this is my chance to convey something about the world the character lives in. Is the character wearing a fresh uniform, or is it bloodstained and dirty? Is it constructed from expensive materials, or poorly made? Has the steel oxidized, or has it been looked after and oiled?
It’s worth mentioning that when texturing assets for games there are certain material types that will be common across all assets, such as skin, metal, leather etc. It’s helpful if material presets have been created early on in the project, to be used by artists across all assets. Along with speeding up the process, it will also lead to a more consistent style and texture values across all characters.
For more details on the character pipeline, I’d recommend having a look at the Polycount wiki.
Personal Approach to Sculpting in ZBrush
During the early stages of a sculpt it’s helpful for me to think about it like drawing. This prevents me from getting lost in the surface detail early on, as my focus is on creating strong shapes and silhouette from different angles. I’ll even go as far as etching lines on my sculpt sometimes, to help indicate major plane changes.
Working this way, focusing on creating a strong primary form, gives me a solid foundation to start building smaller secondary forms on. Once those are established, final detail sculpting can indeed be like a painting, as I’m just creating interesting shadow patterns with tertiary details like veins, muscle striations, wrinkles etc. That stuff takes time, and I think treating it like painting is maybe a useful heuristic. The danger is that much of your detail may read poorly from certain angles and lighting situations if you haven’t taken the time to create strong primary and secondary form. Some sculptors who I think do a great job of creating a beautiful form are Gio Nakpil, Samuel Poirier, and Steve Lord.
I think being able to give yourself feedback is also critical. I do this with paint-overs in Photoshop (good example here by Jon Troy Nickel), sketching ideas on paper and by testing the model under different light conditions in something like Marmoset Toolbag.
Lily Slash Production: Defining Shapes
I have to thank amazing Sergi Brosa for the shapes on Lily Slash. In fact, the way he designs shapes is what led me to do this piece. He has a great grasp of perspective, structure, and form as well, and the shapes translated easily into 3D. I feel like I learned something about shape design by doing this master study!
I started by trying to match the body type and gesture of the character. The original concept art actually had very little perspective distortion, so I was able to use that as a guide for the proportion. Once I had that figured out, I started adding the major clothing and gear elements, always being careful to complement and even exaggerate the shape of the body underneath. The process for the gear was pretty simple, I actually modeled that in 3ds Max. I would extract polygons from the body to get the basic shape and then add thickness.
There was a good amount of detail and complexity on this character, so it was important to block everything out first. The small detail falls into place nicely when working this way, as it’s easier to place the detail when you’re doing it in relation to shapes that have already been placed correctly.
Once I was happy with most of the modeling, I started posing her. This was the most challenging part for me. Luckily, in this case, the pose was essentially a contrapposto, which was a concept I got understanding of from figure drawing. It’s a pose where most of the weight is on one foot, and you get a nice effect where the tilt of the hips is opposing the tilt of the shoulders.
I did Special Delivery speed sculpt over the course of 4 weeks, working in the evenings after work and on weekends. It’s hard to put an exact time on it because of this. The idea came from a quick concept done by a friend of mine, ClouedMeakLin.
I optimized my workflow to get this done quickly, as my goal was simply to create a render and not present a sculpted piece. I had the model lit and textured before the sculpting was even complete, just to see how well everything was holding up. I also tried getting away with using shaders as opposed to modeling detail where possible. A good example of this is her shirt, which has very simple sculpted detail, but the render has a nice complexity to it thanks to the shader. I used a combination of custom blended shaders and VRscans. These are special shaders created by ChaosGroup, that are based on highly detailed scans of real-world objects. They have some limitations but can give pretty incredible results. I would usually blend these in a shader that adds some procedural dust/dirt effects. Using shaders in this way instead of texturing everything was a huge time saver on this project.
VRscans can be downloaded here, along with a trial version of VrayNext.
Video Game Characters Optimization
There are quite many peculiarities when it comes to optimizing game characters. I also find that good information on this is harder to come by, it helps if you have an interest in game engines and the process of getting your own stuff into them of course. Low polycounts and texture resolutions are the biggest challenge, and there are many tricks to get around this. The best way to learn is to study models created by the pros, such as those I linked to earlier
For retopo, I use a combination of Topogun and 3ds Max, often I’ll have created a base mesh when sculpting which can also be easily optimized. In Total War: Warhammer, a character may be limited to as little as 12 thousand triangles. It’s good practice to spend those valuable triangles on important areas like the head, upper chest, hands and joint areas. Generally, there would be a fall-off in detail as you get towards the bottom of the character. Another useful technique is combining elements from your sculpt into one model when doing the retopo. There are some cases where you wouldn’t want to do this, for example, if the parts need to slide over each other or move independently, as may be the case in a joint area.
For UVs I use 3ds Max, with miauu’s Script Pack which has some useful tools for quick seam placement and unwrapping. RizomUV Virtual Spaces is something I’d like to try, it may be a good alternative. When doing UVs it’s generally good to keep an even texture density across the entire model, but that being said the head tends to get more resolution. Areas that are unlikely to be seen by the player can also get a lower resolution, such as the bottom of a shoe. It’s common practice to mirror or stack UVs as well, for example, if you have repeating elements like spikes it’s best to only do a few unique variations and clone the rest.
Depending on the project I use Substance Painter or Marmoset Toolbag and Quixel Suite for texturing and baking. There’s plenty of information on how to go about this online, so I won’t go into too much detail. Check these YouTube channels to learn more:
I’m pretty happy with V-Ray and Toolbag as my go-to render solutions. Toolbag can do more than game assets of course, and I’ll often throw ZBrush sculpts in there to get an idea of what’s not working on a character. It can’t compete with the rendering power of something like V-Ray of course, which is what I use for a final render if I’m not working on something real-time.
I find that V-Ray gives great results even with a simple setup. There is a great support for GPU rendering, meaning I can tweak lights and camera settings in real-time. I really love the look of V-Ray renders, combine that with a powerful shader like alSurface (which I like to use for skin) and you get very believable results with little effort. A good light and camera setup also goes a long way. I’ve always been interested in photography and cinematography, and I frequently steal ideas from both fields when working on renders.
For the most recent character, I used a very common lighting approach, one key light, a rim and some ambient light that’s bouncing from the environment. You can use HDRI lighting for this as well, but in this case, the GI was enough. I did add an extra spotlight on the face, just to help boost the highlight in the eye, and to focus attention on that area.
The hair was surprisingly easy once I figured out the best approach, in this case, XGen and V-Ray’s amazing hair shader. The hair was groomed with splines in XGen and then converted to a very dense polygon geometry. I tweaked this slightly in ZBrush, by using topological masking to pull out some fly aways, and further shape the hair. After I was happy with the mesh, it was simply a case of dropping the V-Ray hair shader on it and hitting render. I’ve also used a warm rim-light on the hair, and a cool shadow, which I think gives a nice effect.
When doing the hair I tried to be aware of how the strands were leading the viewer’s eye, I’m really inspired by how artists like Loish handle hair in this way. The direction of the hair flow is never random and is always done in a way that guides the eye. In this case, the intention was to guide the viewer’s eye back to the character’s face.
To learn more about V-Ray lighting and rendering, I would have a look at anything Grant Warwick‘s done. He has a few podcasts on YouTube as well, which have been immensely useful to me.
As for rendering more stylized characters, I’ll use Football Girl as an example, here’s a breakdown of the scene:
I think most people realize the importance of a good portfolio, and that you will get hired based on the style and type of projects you’re doing. Following your interest is also common but great advice, having a good idea of what it is you actually enjoy working on is pretty useful, and will make studying less of a chore. Decide which artists and studios you admire, and deconstruct what it is that you need to learn to achieve that level, as this will allow for more deliberate practice. For example, if you want to work at Blizzard, a good understanding of anatomy, but simplified in a graphic way, is crucial. Also a good understanding of stylized proportion, gesture and creating iconic, readable silhouette. Once you have broken it down like that you will have some direction for your studies and personal projects. Also, personal projects, do those!
Along with that deliberate practice, don’t neglect the life-long study of fundamentals like anatomy, drawing, and design. Some resources that have helped me are:
- Stan Prokopenko‘s YouTube channel
- Figure Drawing: Design and Invention by Michael Hampton
- Figure Drawing for Artists by Steve Huston
- Anatomy Lessons from the Great Masters by Robert Beverly Hale
- Scot Eaton’s online workshops on anatomy
- New Masters Academy
Another thing that I have found immensely helpful is attending art festivals like THU, Promised Land, IFCC, Industry Workshops, etc. Personally, I think money spent on attending an event like this is an investment that will give a greater return than most art-related degrees.