Luminous Productions' developers took the stage at CEDEC2020 (Computer Entertainment Developers Conference 2020), which was held online in September, 2020. Opening with this 1st interview, we will be releasing a series of speaker interviews on an irregular basis to reveal additional details along with the speakers' thoughts and sentiments behind the development.
The three speakers who gave the presentation titled [Filling an Open World in a Natural yet Intentional Way - LUMINOUS ENGINE's Approach to Procedural Generation-]
－－The theme of your presentation was the "World Editor" - could you tell us about this tool?
Iwasaki ：With the World Editor we focused on how we could have objects placed automatically, yet tastefully, while also retaining the capability to efficiently edit the vast world map for an open world, which can be overwhelming to create. It allows for the artist's unique touch to be infused into the otherwise simple and automatic object placement. So, we aimed to create something that enables automated and sophisticated placement that wouldn't pose limitations to the artist or game designer's creative vision.
－－ So in other words, the tool was accomplished through a collective effort of artists and programmers. The project seems to be a product of cross-professional collaboration - what were the specific roles each of you played in the course of development?
Iwasaki ：I handled the low-level development, or perhaps I should say, the framework around the core part that runs auto-generation. Simply put, I designed and implemented the framework of the procedural generation on the run-time side.
Hamano ：I, as the programmer, was in charge of the workflow and how the editor should actually be used.
Yamamoto ： I'm an environment artist, so I worked on the editor, created by Iwasaki and Hamano, from the user side. I tested the editor to see what was necessary when creating a high quality procedural background and built the network based on that. Also, there were times when I would make feature requests from the two, in order to achieve the artistic expression I had in mind.
Hamano ：Yes, I remember that. It was my first time working with an artist on a tool, so the experience was particularly exciting. It felt like we were paving the way to something new, so to speak.
－－Was the work environment among your team open in the sense that you can exchange ideas and make requests freely beyond your areas of expertise?
Iwasaki ：Yes, it was like we were playing catch with Yamamoto at all times - Yamamoto would tell us "I want to grab this kind of information" or "I need a node like this" and we would make suggestions in response. There were times when Hamano and I had no idea how Yamamoto intended on using certain features he had requested, and once we had them ready, he would use them in a completely unexpected way.
Yamamoto ：We ended up becoming like a World Editor Team that goes beyond our roles and typical responsibilities. Of course, certain tasks were allocated based on our respective specialty, but that didn't mean that each of us would only mind our own business and ignore the rest. On my part, I shared with the team whenever there were bugs or issues raised by other artists, or checked with them about processing load rather than just simply focusing on building networks. So, it really felt like we were putting a collective effort into exploring optimal solutions.
Hamano ：The fact that we were able to work beyond our roles and communicate with each other smoothly helped us with updating the tool in a flexible manner. 。
Yamamoto ：I felt like I was to blame whenever my fellow artists reported bugs. (laugh)
Iwasaki ： It helped so much, too, that Yamamoto is such a tech-savvy artist so he could go very in-depth into this all, allowing us to collaborate on full throttle.
From micro (hand-made) to macro (automation) - [World Editor] enhances efficiency with simplicity
－－ What is the biggest difference between the current World Editor and its past iterations when it comes to developing open world games?
Iwasaki ：By far, auto-generation. With the previous specifications, we could "paint (the map) quickly by hand," so in other words you could choose where you want to paint a tree, for example, then the tool will paint a tree there for you, but it still required someone with an intention to choose exactly what to be painted, whether it's a stone or tree or something else. All you need to do with the current World Editor is to paint thinking "I want a blue-ish forest somewhere around here," and the tool automatically places the rocks, flowers, and trees in a nice and balanced way for you. So, this is definitely a step up in terms of macro editing.
－－It really does sound like anyone would be able to use it regardless of their areas of expertise.
Iwasaki ：Yes, with this tool, anyone can just press a button and create something plausible. The artist should no longer need to spend their time wondering whether a tree should be placed a little more to the right or left and so forth.
Yamamoto ：We might still make minor tweaks like trimming a tree here and there in the final stages of production, but using the current tool as is would suffice at least from the beginning to mid phases of development where things require less precision.
－－Is the tool capable of painting different types of environments, such as cold or warm climates that are quite often seen in video games, in addition to the aforementioned "blue-ish forest"?
Iwasaki ：In such a case, since the types of vegetation and attributes would be profoundly different in cold and warm climates, Yamamoto would first establish a set of rules for each area, then the level designer would take these rules and determine where to apply them. If the differences between two areas are at micro-level instead, then the same rule set can be applied to both. That being said, the difference between a cool region and a desert, for example, would be too drastic to just use the same rules with different parameters, so we'd need to change the entire set of rules.
Yamamoto ： For example, we can establish separate sets of rules for a desert network and a snow mountain network respectively, and create gradation by layering the two. It can be applied in any variety of ways depending on how you look at it.
－－So, you can pretty much create anything as long as you have solid rules established. That seems ideal for an efficient mass production.
Iwasaki ： Yes, it is. In the past, if we had created one beautiful forest and wanted to create another with the same quality, we had no choice, but to plow through the whole process again. With our new tool, however, once Yamamoto creates a set of rules for one decent forest, another forest with the same quality - no matter how expansive you want to make it - can be mass produced without requiring human intervention and without it looking like a carbon copy, although it would take a longer time to compute.
Hamano ：Incidentally, such rules used to be blackboxed and were something only us, the programmers, could tweak in the past. Now that they are networks that can be edited by the artists themselves, they can specify more tasteful and sophisticated visuals with further precision.
－－ You made many customizations to the tool and used those as the base to create new sets of rules to mass produce - it literally is very simple and efficient, isn't it?
Iwasaki ：Needless to say, there are many features out there that are similar to our tool, but what separates ours from the existing features is that we pursued real-time performance. So, basically we customized our tool in a variety of ways and ended up with our "World Editor."
Yamamoto ： From the user's stand point, the real-time aspect is extremely important. Being able to make minor parameter changes while seeing the outcome of those changes in real-time, with the actual models and lighting, makes working on the imagery a whole lot easier. Back when it was blackboxed there was a bit of a speed bump in the process, as the artist was required to send a request to the programmer and the programmer would then implement what the artist needed. With this tool, however, the artist can now directly edit the networks and check the outcome in real-time, so this really allows us to find the correct answer easily and deliver quality within a super quick span of time.
Iwasaki ：We're not intending to do everything on our own without using existing tools, but we're ready to dive in and get the simple processing that doesn't require complicated calculations done on our end, like getting the editing results in real-time for a game.
Hamano ：When you use other existing tools, there are learning costs to think about as well. So, my wish was to enable the users to complete their tasks with simpler operation and making it so that everything could be done just using our own engine made that wish achievable. I aimed for simpler and easier operation with the users in mind.
－－ Has such simple operation made it possible for a base terrain of a certain quality and of varying world view designs to be created, by not just Yamamoto, but other artists as well?
Iwasaki ：Yes, I'd say so. In addition to the preset of "somewhat natural" objects that can be used for any types of games, you can add a flavor that is as unique as you want to your own game.
－－The sample terrain was very realistic, but could this also be applied to a cartoon or cel-shading type of approach?
Yamamoto ：Yes, expressions like those are fully achievable depending on how you build the network.
Iwasaki ： Oh, and one more thing. "Just press a button to get something entirely auto-generated" is our most procedural-oriented approach, but it can be made to lean more toward hand-crafted as well, where you specify by hand what you want and where and the auto-generation follows those specifications. So, let's say you want to add a bit of hand-made flavor like "there should be undergrowth at the base of a tree" or "when a curve is drawn, the surrounding terrain should get slightly dented," you can dictate that and our tool can take over from there - this is what we're referring to when we say "micro to macro." You could use a different tool to add a bit of hand-made flavor, but using another tool just for that purpose is a bit much and leads to a drop in iteration speed. So it's ultimately more efficient to have the same tool cover everything.
Yamamoto ：Additionally, letting the artists build networks themselves really makes it possible for their artistic sense to come through. When you ask the programmer to implement something and it comes back not entirely what you expected, you go through a phase where you have to try and describe your "artistic sense" to the programmer with words, which really is quite impossible. If you can build the network yourself, you can express what's in your imagination. So, while you have to learn how to use the tool, it can still makes things easier for the artist as they can get the positive outcome they're looking for right away.
Why make it? Because we can! - Creating anew to take a step toward innovation
－－So the team takes advantage of existing tools and the in-house engine where appropriate, while upgrading and enhancing the in-house engine on a continuous basis to meet the studio's needs?
Iwasaki ： I might sound like a showoff, but whenever there's a certain feature we want, we search to see if it exists in the market. If it doesn't, rather than modifying our workflows to fit into an existing tool, we'd create the tool ourselves - that's what Luminous Productions is about. We don't just piggyback on what already exists and hope for the best with it, that's not who we are. The mindset most of us here have is that, if it doesn't exist, we'd go as far as to create it ourselves as long as it's guaranteed to help us achieve the quality we need. I mean, every tool that is available in the market now is a product of someone's quest for something new. So, the idea of creating something new to take a step towards innovation is important to us.
Yamamoto ：With an external engine, if a feature we wanted didn't exist, we'd simply have to give up. However, our proprietary engine, while it does cost us time and effort, gives us the environment in which we can get what we want, and I'm very grateful for that as an artist. Even a simple request like "please fix this UI as it's hard to use" can be fulfilled, too.
Iwasaki ：From an engineer perspective, I think, the moment we stop developing new technologies, we'd get left behind in the long run. For example, I won't mention their name, but the reason one of the car manufacturers here in Japan grew to be a major corporation was because they were determined to create their own cars, rather than throw up their hands and go for existing imported cars instead. So, why do we create rather than just use what already exists? Because there are greater results that await us there. If we choose to go for an existing engine, our growth ends there. If we want to be a leading company in the industry and continue to be at the forefront of technology, I don't think using existing tools would be enough. We have to keep inventing on our own.
Hamano ：The World Editor was created from the ground up and it can still be enhanced in infinite ways from here on out. And it is that possibility of enhancement that gives us the options we can only get from using the Luminous Engine. Our ambitious development effort rewarded us with this solid foundation that can still be taken anywhere.
－－What you just said about the engine, can also be said about Luminous Productions itself can't it - that we're challenging and continuing to upgrade ourselves on a continuous basis, right?
Iwasaki ：Yes, we're constantly challenging ourselves. I find the act of creating somewhat romantic so to speak. (laugh) In a general sense, we put great emphasis on creating things ourselves.
Yamamoto ：I agree. We have reached the ideal of having the ability to materialize our wishes. We're always on a quest for new approaches. We are a team of high-fliers capable of working toward realizing the ideal that everyone shares.
[Luminous Productions] and the [Luminous Engine] are continuously being upgraded by creators who never cease to challenge themselves to create something new, to stay one step ahead of the world.