Martin Holmberg and Thiago Klafke from Blizzard Entertainment talk about the tools and their design decisions that helped to shape Temple of Utu – an incredible environment which became one of the finalists in the Allegorythmic Throne Room challenge.
Martin Holmberg and Thiago Klafke from Blizzard Entertainment talk about the tools and design decisions that helped to shape Temple of Utu – an incredible environment which became one of the finalists in the Allegorythmic Throne Room challenge.
From January 14th to March 31st Allegorythmic and Polycount conducted a The Throne Room contest. It was an environment art challenge that encouraged artists from all over the world to create fantastic scenes from imaginary virtual kingdoms. There were a lot of works submitted to the contest. Among those was Temple of Utu, which became one of the finalists in this competition (2nd place). This project was created by Martin Holmberg and Thiago Klafke – two Environment Artists for Blizzard Entertainment. They were very kind to share their thoughts about the development of this project and the tools you’ve used during the development.
About the Developers
Martin Holmberg is a Senior Environment Artist for Blizzard Entertainment, where he has worked for the last 5 years. He has been involved with Titan, Starcraft II Legacy of the Void, and will be joining the Overwatch team in the near future.
Thiago Klafke is an Environment Artist for Blizzard Entertainment and has been with the company for 3.5 years. During this time he has worked on Titan, as well as Starcraft II Legacy of the Void.
The Creation of Temple of Utu Backstory
When we set out to make this scene, we wanted to create something inviting and enjoyable that would engage the viewer to want to explore the scene. The next step in the process was to build a backstory for our project that we would be both excited about making, and would achieve the atmosphere we wanted to create. At this stage it is important to have a fairly uncomplicated story that is easy to grasp for the reader. A simple idea with only a few key elements is also a lot easier to stick to while making your scene. We can’t stress how important it is for everyone on the team to be excited about what you’re making. Unless everyone is 100% onboard, you’re off to a rocky start.
It is also very easy to fall into the trap of making an overly detailed backstory. The result of this will probably be that most of your viewers won’t get what you’re trying to tell them through your art due to the complexity of your storytelling. Secondly, you will have too many elements in your scene and it will end up being hard to read and be overall incohesive.
Once we agreed on the backstory and mood, we gathered some material reference that we wanted to use. This might be personal preference for us, but we feel that when gathering ref, less is more. Stick to a few images that capture the look you want to go for. Too many ref images will only muddle your original idea. Keep it simple! It’s also important to keep the number of your materials relatively low for a number of reasons, the main being readability. Too many different materials will create visual noise, and make it confusing for the viewer/player. For example, we decided on sandstone and marble for our main materials, and for trims we used mainly stone and dark gold. A lot of secondary color keys and detail came from the props and vegetation we did. The banners, palm trees, and statues did a lot of work to make the environment look lived in, and provided a breakup in both our color palette and modularity.
Tools Which Built The Temple of Utu
The tools we used to create the assets for the scene were 3D Studio Max, Quixel Suite (NDO,DDO), Zbrush, 3D Coat, XNormal and Photoshop.
3D Studio Max is a really sturdy allround 3D software. It has great modeling tools, and the modifier stack it uses is really versatile and easy to get into. Also, since it’s such a popular software, there’s a ton of documentation and tutorials online that will help you if you ever get stuck on a problem.
XNormal is pretty much the only texture baking software we use. It’s really accurate and easy to use and we love it. We used Zbrush to sculpt our trim textures, statues, and the throne itself. We used Photoshop to do texture painting and composite the bakes we got from XNormal.
The engine we chose to assemble the scene in was UE4.
Epic made a huge leap forward with this latest installment of their engine and it proved to be an amazing tool to use for our project. Sending materials, assets and entire project files back and forth between us was easy, fast and unproblematic. The documentation provided with UE4 was also incredibly helpful and easy to understand. It has an amazing material editor. Your shaders are really as complex as you want them to be, and the node based creation process makes assembling shaders a very visual experience that’s easy to grasp and follow.
It’s also worth noting that Martin had next to no experience with UE4 when we started this project, but learned along the way with the help of their online documentation and video tutorials. While he feels he has a solid understanding of the engine, he now knows he only skimmed the surface of what the engine is capable of.
One tool we wished we had time to learn was of course Substance Painter and Substance Designer. We’ve seen some really cool materials being made in Substance Designer and now that the challenge is over, can finally sit down and learn the software properly. We think this program has a promising future in a games industry pipeline for a number of reasons, with speed and consistency being the biggest factors. One issue with large art teams is that everyone paints their textures slightly differently and the way people structure their Photoshop layers might make it next to impossible for other artists to go in and make changes. With Substance, a team could set up a material library where all your material values would be established, and editing these master materials would be much easier and lead to more consistency stone, gold, brass etc. would look the same throughout the game or scene.
Speed is another factor. Painting a 1024 or 2048 texture by hand is very time consuming. Substance Designer will let you generate high res, accurate textures quickly and efficiently.
Open Environment vs Closed Environment
Thiago: This is all very relative, but I think that broadly speaking the main differences are the scale and the relationship between the elements. When working on an interior scene you want to create a sense of organization and cohesion, the space has to look functional and the details need to hold up at close distance.
When working on open world environments, however, the most important aspect becomes creating a sense of distance and space. You have to use a different set of “tools” to make sure the read of the scene is good. On open world environments your skybox is also one of the main “characters,” so make sure you have something that looks inspiring and conveys a lot of information of what you are trying to achieve.
I think an environment really shines when both aspects are combined in a seamless way. When working on huge environments you still need small scale props to sell the scale. If it’s a city you will need props that people can relate to such as storefronts, doors with the correct knob height, benches, etc. Make sure the scale on these props is perfect. People can ignore a few windows not having the correct height on a building, but if a chair looks bigger than it should it really throws everything off.
On the other hand if you are working on an interior scene it really pays off to invest in a cool backdrop. Sometimes you don’t need much – remember the space level in Duke Nukem 3D and how awesome it was to look outside the window and see the space shuttle and the Earth?
I think a game that does this wonderfully is Final Fantasy XII. The playable areas are extremely small but the transition to the backdrop is pretty much flawless so everything look much bigger than they really are.
I personally feel both closed environments and open world environments present the same types of challenges, the difference is how you tackle them depending on your scene. Always establish a point of interest that is the centerpiece for your environment. It can be something as simple as a light source or something much more complicated like a firebreathing dinosaur with fricking laserbeams for eyes. Whatever it is, let it guide the player/viewer to where you want their attention to go. Sightlines and unaccessible areas are also great ways to give the illusion of space.
Our throne room uses a few of these concepts to make our scene look massive. The balcony looking out on our background buildings was a great way to give the player the idea that the temple is in the middle of a bustling city. What the viewer doesn’t see, she will fill in with her imagination, which is a really powerful tool. All you need to do is put the idea in her head that there is something there and her imagination will do the rest.
Same thing goes for the lit doorway to the left or our scene where a lot of sunlight comes through. It could be a balcony looking out over the city, or maybe its a path to a garden, it’s really up to the viewer to decide.
The pillars, banners and vegetation also do a great job at breaking up sightlines in the room and make it feel more dynamic. You should never be able to stand in a game environment and see the whole room around you, then there is no reason to explore. Hide parts of it and make the player interested in exploring what you have created.
Level Design as Part of the Game Story
Martin: I feel level design and environment art have always played a key part in the storytelling of games. The only difference is that today we have awesome engines like UE4 at our disposal which just gives us access to even cooler tech than ever before.
Thiago: I agree with Martin that the new tech really sets us free! In my opinion story heavy experiences will become even more necessary than ever before with the advent of VR. All you need in VR is a nice experience, so games like Dear Esther will translate perfectly to that new media. It’s really a brave new world and for us environment artists it’s a dream coming true because we can finally focus on making kickass environments that feel alive without having to limit our vision because of gameplay constraints. We will finally be able to make “games” (I’d rather call them experiences) that consist only of kickass environments where all you do is just walk around and enjoy your time there.
A Thin Line Between Good and Bad Environment
Martin: Every environment should start off with a great idea that you’re passionate about like we discussed earlier. The next step should be the blockout. I like to do this in 3DS Max, then drop the mesh in UE4. Your scene should look cool as a flat shaded greybox environment. If it doesn’t, no amount of fancy textures and statues will change that. Once I’m happy with the space I’ve blocked out I move on and start blocking in materials and play around with the set pieces to create alternate sightlines. To me, that is good environment design.
Thiago: To complement what Martin said in a more abstract way, to me what makes an environment work is if it feels like a real place. Sometimes there’s this misconception that a game environment only works if the art is awesome and everything is perfect, but I strongly disagree with this. When I play games I want to feel immersed in the world and this happens when everything makes sense. By that I don’t necessarily mean making a realistic city with cars stopping at automated traffic lights, but setting a few “rules” for your environment and sticking with those. The best example I can give to illustrate this is Skyrim. If you look at the individual pieces they are not masterpieces, what makes that game so good is how everything blends nicely together. When you are walking around that world you really feel like you were transported to an alternate universe where virtual people live alongside dragons and magic. This is where your backstory come into play. Ask yourself these simple questions and I guarantee you will always be on the right track. What is the level of technology of your environment? How people live in this place? How well maintained is everything? What time is it set? Is there any major event going on that you can hint at the environment? For example if there’s an alien invasion going on in your city you can fill the walls with anti-alien graffiti or propaganda posters instead of jokes or cool art you like that has nothing to do with that.
A very common mistake when making environments is focusing only on adding shiny cool stuff and letting the vision fall apart. Before adding anything you should ask yourself: is this going to improve the cohesion of my environment? If the answer is no, move on to other things.
I’m very proud of what we accomplished with our Throne Room because that’s what we set out to do when we started. I think the best compliment an environment artist can get is when people say things like “I want to explore that place” or “wow, it feels like a real place”.
The Long Awaited CGI-quality in Games
Martin: I feel game environments still have a long way to go in terms of hitting the same fidelity as movies. The main reason though I think is time. Movies never have to worry about what’s outside of the camera’s field of view. We as game developers do. So in short, the problem is the volume of work you have to do. The good thing is that software like Substance Designer lets you make high res textures fast and clean, and UE4 does a great job at rendering it all out for you. It’s really just up to us as environment artists to work smart and invent new pipelines that use the full benefits of these programs.
Thiago: Yeah, I agree with Martin. On one hand it’s good to still have limitations though. We artists tend to go really overboard with details and stuff we want to add if there’s no one watching us. Imagine how long games would take to be made if we could place an infinite amount of art in a level.
We hope this article sheds some light on how we built this environment and that the reader found some bits of useful information that might help them in their future projects. Our sincere thanks also go out to Allegorithmic for hosting the awesome Throne challenge. We learned a ton from it and can’t wait to explore Substance Designer further.
Martin Holmberg and Thiago Klafke