Experiences environment artist Jobye Karmaker shared some of his thoughts about vegetation creation in video games and described how to turn the environment into a gameplay tool.
We were fortunate enough to chat with Jobye Karmaker – an experienced environment artist, who worked in some of the biggest game studios in the world. He talked about the way you can make the virtual world a part of general gameplay experience and described his approach to vegetation creation.
My name’s Jobye Karmaker, I’m originally from Montreal, Quebec, up in Canada. I’ve just recently moved to Seattle, Washington in the United States to work at Monolith Productions as a Senior World Artist. Before that I worked nearly 5 years at Ubisoft Toronto as a Senior Level Artist where I had the chance to work on a variety of both singleplayer and multiplayer games in multiple genres: Rainbow Six: Patriots (what eventually became Siege), Splinter Cell: Blacklist, Far Cry 4 and most recently, Far Cry Primal. Previous to that, I worked for several months for Simthetiq making content for serious games and military simulations. Before working professionally in games, I used to work in the Source Engine modding scene and contributed art to WW1: Source & Ham and Jam, which was super exciting at the time given that I was a big fan of Day of Defeat and Half-Life 2.
How I broke into the industry, it was a combination of two things for me. The first of which was going to school. I attended Collège de Bois-de-Boulogne for 3 years, where I received a Technical Diploma for 3D Animation and Modeling. Also, I attended the NAD Centre, which I dropped out of after one semester due to landing the job at Ubisoft. I didn’t go the self-taught route given that I felt I needed the school structure at the time, though I did a fair bit of self-learning in my last year of school to make sure I landed a job at the end of my degree.
Second, where I actually found out about the open Modeler position at Ubisoft Toronto, was on Twitter, before it was even posted on Ubisoft’s job site. Social media and socializing with industry people on there can be very important for both your career and even personally. I ended up meeting and becoming friends with a lot of people I had met on Twitter before I met them in real life when I moved to a new city.
The Creation of the Environments
Anytime I’m tasked with heading up an environment, there’s usually already a high-level brief from the Art Director and sometimes a piece of high-level concept art to get you going. From there, for me, after breaking down the concept and sourcing reference, it’s important to nail the feel (and not necessarily 100% of all the details in it) of that high-level direction and concept while also respecting and adding to the gameplay along with the Level Designer. It’s only ever a jumping off point for the artist. It’s up to you to break it down with your Level Designer and sort out exactly how you’ll build it out. From there, as Level Artists, it’s super important for ‘the big picture’ to be nailed as early as possible. Once you have that base, any detail you add on top of that is just gravy and just keeps making that big picture better and better. Past that, I believe it’s important for any environment to be grounded, to reflect its context and story of those who’ve lived or are living there. Everything should be placed with purpose and never just there ‘to be cool’. This is a topic I’ve covered pretty in-depth in one of my blog posts about some of my guiding art principles which was linked on 80.lv.
Seen above, an area where we nailed the ‘Big Picture’ fairly early on and constantly iterated on top of this main approach.
Environments as Part of the Gameplay Tools
Oh definitely. Some games do such a great job of it that reading the environmental queues almost becomes second nature. Rise of the Tomb Raider comes to mind recently, especially in their escape/chase scenes where you rarely hesitate on where and what to do. First and foremost, I feel environments need to flow well (e.g. not getting hung up on random small collisions) and players should never feel impeded by it in ways that feel unjustified (e.g. little wood barriers that look jump-able should never block out-of-bounds areas!). Second, I think the best way to guide the player is by contrast, whether it be through lighting, texture, model placement, etc. You see it very clearly in games like Tomb Raider and Uncharted with their contrasted climb ledges for example. In more open-ended games, we always try to catch the corner of your eye with those techniques, as you’re walking or driving along a path, to help lead off you off the beaten path and lead you to discover new locations and new points of interest. Fallout 4 is phenomenal example of that, I always get lost exploring that world for hours without even touching the main quest. For some additional watching, Game Maker’s Toolkit has covered this topic pretty well:
First and foremost, I think vegetation brings incredible life and movement to an environment, especially when it reacts dynamically to player actions, the wind, etc. Given that games like Far Cry rely on vegetation so heavily and make up almost 60-70% of the screen real estate, vegetation most definitely plays a large role in helping establish mood and atmosphere for any given area. In Far Cry Primal, the environments that each of the 3 tribes inhabited definitely reflected them. The Udam’s Northern Arctic Tundra area for example stands in stark contrast versus the rest of the game’s lusher areas. Shapes and silhouettes are more aggressive, colors less vibrant, etc.
Seen above, the Udam tribe’s Tundra environment reflects their character.
We used a pretty SpeedTree-centric pipeline. We’d initially tightly model (to reduce dead pixels) the smaller parts we need such as branches and leaves card in 3DSMax, then we’d bring those in SpeedTree to create 3D branch clusters (rather than use a single card with baked down leaves and branches), which we would then bring into a main trunk & branches file to instantiate on, and then, from there we could get that in-engine. For modeling, we didn’t use Zbrush too heavily apart from large shape definition when making custom trunks as we’re finalizing our trees.
As for material production, we very honestly kept it fairly simple. Our Texture Artists used heavy photo-sourcing (using scientific names of trees helps great to find nice high-res images!) and leveraged nDo2 and CrazyBump masterfully to create our barks, leaves and branches.
If you’ve authored your vegetation with both branches and leaves being separate cards, it makes it relatively painless process to make ‘dead’ variants of some existing trees from the lusher biomes to populate snow climates (just taking out the leaf cards!). Along with leveraging the AO bake system from SpeedTree, we were able to use that as a mask and add a layer of snow or moss on the trunks depending on the biome a tree would be used in.
Seen above, dead variants of trees that previously had leaves in other biomes.
First and foremost, it’s important to gather solid references about specific tree species you’d like to make: breaking it down, studying how it grows, how it’s composed, etc. As mentioned earlier, if you’ll be heavily photo-sourcing your textures, it’s worth using the scientific names of trees to get some nice high-res images.
Second, I don’t think you need a ton of variants of the same tree to achieve variety. I think you’re better served having less variants of each tree species and focusing on building it smartly. You should make sure that almost every angle of the same tree feels like a new tree by paying attention to the shape and silhouette and not making it too uniform. That way, you can focus on having more species. (e.g. if you need to build 6 trees, doing 2 smartly built variant multiplied by 3 species gets you a longer way rather than 2 species with 3 variants of each.)
Third, as much as possible, if you’re building multiple biomes, make sure to share some trees between each of them to help them blend into each other better so it feels like a nice gradual transition as you’re walking through the world.
Seen above, more species, less variants of each, gets you a long way.
What do you think are the best kinds of tools if you are approaching environment modeling & texturing? It would be great if you could give some recommendations to people starting out.
These days, for people starting out, there are so many different ways to approach modeling and texturing and so many different tools to use (3DSMax vs Maya, dDo vs Painter, Zbrush vs Mudbox, etc.). At the end of the day, tools are tools, given the right time, it’s fairly smooth to transition from to the other. The skills you develop in one are not necessarily lost in another. So, experiment, see what you like and are more comfortable with and focus on that set of tools (e.g. learning Zbrush/Maya/Painter vs Max/Mudbox/dDo). You don’t need to necessarily learn it all, but know one set at least. If a particular job uses another set of tools, you’ll usually be given enough time to ramp up on them properly, don’t shy away from one because it’s not the one you use.
That said, there are so many great resources online nowadays to learn from. Polycount is of course a staple, Gumroad has a wealth of free and affordable paid tutorials straight from all the greats of the industry and not to mention all the other paid tutorial sites.
As always, if you ever have any questions, I’m always open to people reaching out to me, catch me on e-mail or @jobyek on Twitter. You can also see the rest of my work on my website.