3d artist and educator Jeremy Estrellado discussed his approach to building virtual spaces and shared some advice on the best way to create playable environments.
My name is Jeremy Estrellado and I’m a Senior Environment artist at Massive Entertainment, Ubisoft in Malmö Sweden. I work on The Division as you might have seen in a previous 80.lv post. My contributions on that project mainly center around the Base of Operations and the Terminal. In my spare time, I run a stream that’s focused on learning what it means to be an environment artist in the games industry as well as relevant discussions. I really like to show that even with lots of experience, you will always be learning, which is why my motto is: “Creating better art with you.” You are all welcome to visit at Twitch or check out the archive here. One of the most valuable parts of my stream that everyone seems to look forward to are the critiques at the end. I pull from my discord community work in progress section which Lasts about an hour.
My interest of making art for games started quite young, around the age of thirteen. I have two uncles I look up to who were also heading in the same direction. I went to The Art Institute of Seattle, which was one of the few schools teaching about making art for video games. Around the three-year portion of my schooling, I was offered an internship to work on Pirates of the Burning Sea with Flying Lab Software. Approaching graduation, I posted about my future goals on the forum Polycount. There were a couple of people commenting on my portfolio and asked if I was going to be at the onsite portfolio review that the school offers near graduation. I later found out I was talking with a representative from Zombie Studios who visits the school reviews every year. They offered me a position as a Junior Environment Artist later that week. That’s how I got my first full-time job.
At Zombie Studios I started on SAW the Video game under the mentorship of Steve Holt. From there I went onto working at Snowblind Studios, who were just starting full production of Lord of the Rings: War in the North. I then had the opportunity to help finish up for launch as well as work on DLC for Gotham City Imposters, a large but fun stylistic departure from my past work. I headed back to Lord of the Rings Franchise with Shadow of Mordor for WB Games, Monolith. Having a need for a change of scenery, I decided to take up a senior position at Turn 10 Studios. Helping usher in the new Xbox One with Forza 5 and helping produce some of the DLC was a very cool experience. It’s not every day you get to support the launch of a console. Near the end of the project, Massive Entertainment approached me about coming to work for them in Sweden.
It’s been almost three years since then, but I’m still very happy and excited for all the traveling and experiences my partner and I have had the chance to do so far. It’s amazing and I’m happy to have the opportunity to be working on The Division with our amazing CoDevs: Red Storm, Reflections, and Annecy. Expect some screenshots on my ArtStation in the near future.
Immersion is everything, no matter which game you’re working on. You want to feel like you’re in the game as much as possible. To be lost in another reality is an amazing feeling that people who play games are looking to do. The biggest takeaway I can give on this is interaction involving the character and the environment surrounding them. Everything from the character being able to grab ledges, stepping on small changes in the height of the ground or when the bullets go flying and everything gets destroyed accordingly. The more you can push the level of interaction between the player and their environment the more you can increase the immersion. As an environment artist that’s the most important thing. I feel myself and others should be focusing on that. It encompasses all types of games.
Keeping your skills strong is really important. The reality is that in my everyday work Zbrush hasn’t been part of the process. Getting things done with deadlines can be difficult when it’s easy to just sit in Zbrush all day making things look cool. everyone on the stream I’ve been using Zbrush because it tends to be a common topic for discussion and questions. What brush to use? How do you do this? I’m having issues with this. How do you get this problem to not happen? Zbrush is also fun, so there is that.
I switched to Modo after using 3ds Max for 5+ years. During school, I was learning Maya. My first job used Maya and 3ds Max. I chose to expand my skills into 3ds Max while I was there. During my time at Turn 10 Studios, we were seeing videos of Modo scripts. It quickly became something I was interested in learning. I switched when I got to Massive since we support Maya, 3ds Max, and Modo. With the help of my lead, I was able to smoothly switch on the job over the course of about three weeks.
In my current scene, I’m chipping old stonework for a temple scene based off an old Uncharted 2 concept by John Park. The chipping I do with a slightly modified trim smooth brush and the ornate details I leave to Modo. I make sure to build a library of those ornate parts so that it becomes easier and faster to make those details. While I stream I like to remind everyone that you can make whatever you want in whatever software you want nowadays. It’s more about what you’re comfortable in to get you into the zone vs using the new tool of the month. Today it’s pretty easy to overcomplicate the process when you could just do it the way you know and get close to if not the same result.
Most recently I started using Zbrush more to make a majority of my low poly’s directly from the high poly. I use a combination of polishing and dynamesh to get my results. The early parts of the process I can thank Ben Wilson for.
When the industry started making the shift to Physically Based Rendering, there was added complexity creating a need to be able to edit all the different maps simultaneously. With the Introduction of things like Substance Designer and Painter as well as The Quixel Suite, it became much more manageable and faster in most cases. I understand that Quixel Suite uses Photoshop but with the introduction of the Megascans Studio, I can see in the long-term Quixel will be building its own standalone Program.
My current workflow revolves around layering materials and masks. I build all my base materials in Substance Designer. Think of them like basic generic materials like stone, moss, or aged copper. I use Substance Painter to generate all the masks. I export my masks using mask export. It’s pretty handy since you can have it export at any resolution on export. (up to 8k) One of the downsides initially with the exporter is that it didn’t pad your masks. This can be resolved by Editing the script. Scroll down to the bottom and you can just set infinite padding to true. I wish everything were that easy…
While I would suggest people don’t attempt to do a scene as big as mine, it has some special benefits for my situation. By picking something so big It gives me something to chew on for the stream. Sure it will take a lot of time but it will also give me tons of problems to solve with the community. Another benefit is the questions and solutions that we come up with end up being a huge educational positive for everyone involved including myself.
For the scene I am currently working on, I feel building modular is the smartest way to approach it as a whole. Working with a large scene allows me to explain and give examples as well as answer questions that anyone may have. To spark discussion and answer questions is the major part of the reason I stream my growth as an artist. Many questions about how to do modular design or what makes a good modular set were asked often early on and ultimately led to me working modularly.
To build a nice asset that’s modular but also sculpted sounds like it could be tricky, but it’s pretty simple thankfully. Blocking out all your modular parts and getting them to align properly in UE4 or Unity is the most important part. Once that’s working you will have the perfect base to build from within your 3d applications as well as in Zbrush. I leave the patterns to be built inside of Modo but Zbrush can work as well.
For more info on modular building check out this tutorial.
The usefulness of an object boils down to how often it is used and what it brings to the scene as a whole. How often also will help in deciding how much work and detail needs to go into a given part of a modular set. Plan it all out first with block out meshes and then move forward with where details should be. You wouldn’t want to work on high details for a wall piece only to learn that it is thirty meters above the player.
Adding details is a great way to make something look awesome, but also a great way to become noisy and hard to look at. Compositionally speaking it’s best to make sure that if 30% of the asset or scene is high-frequency detail, the other 70% is smooth or easy on the eyes. Areas of visual rest are very important to the overall image.
Usually, I’m now documenting things I discover while streaming or just posting a progress shot. Before I started streaming, I had a habit of wanting to solve a technical problem or getting a process figured out, only to figure it out with the first asset, quickly losing interest and moving on. Streaming has helped me progress further and faster.
I’m currently using UE4 since it offers so much. Its’s so flexible for building shaders and environment art out of the box and for free. My biggest weakness is having to work with lightmaps exclusively. It is something I’ve grown tired of over the years. It’s one of those technical things that just gets in the way when I just want to get to making something look nice.
With some additions being added like Amplify Shader editor, which has recently exited beta, or the popular ShaderForge editor; Unity is getting much closer to the fidelity I expect in a current generation art portfolio. I would argue that it’s already possible. The main reason I really like Unity as a possible option is its ability to do baked and or Real-time global illumination. I still need to play with it more to see where it works and doesn’t work for myself but it’s pretty nice to work with Enlighten vs the old school lightmap solutions.
I have tried Toolbag3 in the past as well as its current iteration but it has become too expensive for me to look at when you have Unity and Unreal sitting there for free waiting to be used. It has however just received some material blending shaders from a third party. While expensive, Toolbag3 is still a very good tool for showcasing single assets or small dioramas.
Open World and Enclosed Spaces
Open world or outdoor environments, as well as indoor or enclosed places, have their own set of problems that must be solved. Some things to think about are: is it co-op or single player? 1st person or 3rd person? is it cover-based? The answer to these questions will drastically change how you do things or approach problems.
Is it co-op 3rd person as well as cover based? Outdoors will require you to make sure there is enough cover for the right amount of players, they need to have cover options. What does adding indoors add to the above scenario? Things like hallways having to be wider to support more players and also include enough cover options. Having a game that transitions back and forth between indoor and outdoor will require a system that allows for the two spaces to meet-up nicely without the player even realizing it.
Details in the environment have key elements that bring meaning to a scene. If there isn’t any meaning to them being there they just add noise for the sake of noise. My suggestion: when dealing with propping a scene, the best approach to detailing an environment is to break up laser lines ‘long continues edges of geometry’ or areas that make the scene look unnaturally empty. These can give a 3d generated look to the scene that can break the immersion of the player. You can break up laser lines by telling a story through propping. Storytelling through props is probably the number one best reason to add details in your environment.
To break a scene down think about it in layers. Buildings and props that are always supposed to be there first. For example, city structural things like buildings and utilities (sewer caps, fire hydrants, signage, etc.) Next layer could be the state of the world. Has the place been looted? Is the place post-apocalyptic or abandoned? The final layer could be the story propping layer. Adding layers in this way can naturally give a lived-in feel to your environment.
The best advice I could give to environment artists starting out is to start small. I know you want to be able to make an entire scene but you will be met with a giant mountain to climb. Start with modeling objects in real life which you can gather unlimited references on. Don’t worry about texturing it or making it low poly, just make it how you see it. This will get your modeling muscles stretched and ready for the work ahead.
After being able to reproduce real life assets from reference, it’s time to get a perfect normal map bake from your high poly to your low poly. This is, in my opinion, the most important thing for current generation game art creation. A note that this part can be pretty rough for people since it can become technical the first few times.
I’m a bit of a nerd and have an excel sheet that I use to plan all of the production out. Going from a block out to the final mesh has stages in the process. This is really only needed when you’re building a large scene that has many assets that you need to keep track of. If you want to try with single assets this is good practice to keep yourself organized early on.
The last thing I would suggest is building hotkeys as well as macros. Lowering mouse clicks as well as taking care of common repetitive tasks will speed up your process.
Thank you for taking the time to read this and I look forward to seeing you on the stream.