Back in February, Amazon announced that they had purchased and forked Crytek’s Cryengine and made the full source code available in a free AAA game engine offering called Lumberyard. The only catch is that if you want to use any public cloud service, then you have to use Amazon Web Service offerings.
At SIGGRAPH, I had a chance to talk with Hao Chen, a senior principle engineer for the Amazon Lumberyard, about Cloud Canvas visual scripting interface to AWS and GameLift multiplayer offerings. We also talked about some of the research and development areas such as integrated artificial intelligence offerings, natural language processing with Alexa, potential ecommerce solutions, and research into digital lightfield capture, compression, and delivery. Amazon wasn’t making any specific new product or gaming content announcements, but it’s clear that part of Amazon’s long-term strategy is to rely upon game developers using their public cloud services in order to fund and sustain future development on their Lumberyard game engine. You can hear more about some of the existing features and functionality of Lumberyard as well as some future research on today’s episode of the Voices of VR podcast.
LISTEN TO THE VOICES OF VR PODCAST
Donate to the Voices of VR Podcast Patreon
Music: Fatality & Summer Trip
Rough Transcript
[00:00:05.412] Kent Bye: The Voices of VR Podcast. My name is Kent Bye, and welcome to the Voices of VR podcast. So back in February of 2016, Amazon bought the Crytek CryEngine. They essentially licensed it to be able to fork it and release it as full source code and to be able to create another AAA game engine that's available for people to use in order to create games as well as immersive virtual reality experiences. And the reason why Amazon did this is the one catch is that if you end up using any public cloud service then they want you to use the Amazon web services. And so they have a number of different services and offerings in terms of doing multiplayer and A lot of future plans when it comes to artificial intelligence and e-commerce and digital light fields and all sorts of different functionality that is really well suited for the cloud. And so I'll be talking today to the senior principal engineer on the Amazon Lumberyard project. His name is Hao Chen, and so we'll be covering what is already out there and available in the beta release and what's to come. So that's what we'll be covering on today's episode of the Voices of VR podcast. But first, a quick word from our sponsor. This is a paid sponsored ad by the Intel Core i7 processor. You might be asking, what's the CPU have to do with VR? Well, it processes all the game logic and multiplayer data, physics simulation and spatialized audio. It also calculates the positional tracking, which is only going to increase as more and more objects are tracked. It also runs all of your other PC apps that you may be running when you're within a virtualized desktop environment. And there's probably a lot of other things that it'll do in VR that we don't even know about yet. So Intel asked me to share my process, which is that I decided to future-proof my PC by selecting the Intel Core i7 processor. So this interview with Hao Chen happened at the SIGGRAPH conference happening in Anaheim, California from July 24th to 28th. So with that, let's go ahead and dive right in.
[00:02:14.926] Hao Chen: My name is Hao Qian. I'm the Senior Principal Engineer for Amazon Lumberyard. And Lumberyard is a AAA quality game engine with full source connected to the cloud. And we are one of the engines that provides a complete solution for VR. And we support Oculus Rift and HTC Vive and with future devices coming.
[00:02:37.282] Kent Bye: And so I understand that Amazon, with the Lumberyard, actually purchased a full copy of CryEngine and then open-sourced the code. So maybe you could talk a bit about how that came about and what that is enabling you.
[00:02:50.998] Hao Chen: So CryEngine has always been one of the high-quality game engines for AAA games. It has built a very strong reputation of the highest graphical quality, good performance, and live editing, meaning that what you see in the editor is what you see in an engine. So we thought that was a really good starting point for us, but obviously we want to go beyond what CryEngine can offer. One of the first things that we did is connect the engine to the cloud, so we offer several really big advantages. in the cloud services, like Gamelift, which allows you to automatically scale session-based multiplayer games with just a few lines of code. We have Cloud Canvas, which allows you to utilize all of the AWS services with a visual scripting language. And again, it adds big convenience to our developer customers. Beyond that, we'll continue to refine the engine and add new features. and work on, you know, forward-looking future stuff to make sure that the game developers get the best engine out there with a full source so that you can modify the engine to suit your needs and also, obviously, have the AWS cloud behind this engine. It's a big, big, huge advantage for developers.
[00:04:04.327] Kent Bye: Now, I recall that Lumberyard was announced at GDC for the first time. Has there been an official launch at GDC or has it officially launched since then?
[00:04:13.149] Hao Chen: Yeah, so we announced Lumberyard, and so we're currently in beta. We have a regular release cadence, we'll release to customers, and we take feedback from customers to make it better. So we're still in beta right now, so there are a lot of features that are still being worked on, and bugs are being fixed, and performance improvements and workflow improvements are constantly made. And then we're adding, you know, exciting features and a constant trickle. So you could say that, you know, Lumberyard is announced, but we'll continue to add improvement until we determine that we're out of beta. But right now our focus is to gather the customer feedback and supporting the existing customers.
[00:04:55.314] Kent Bye: And for someone who's a virtual reality developer, maybe you could go through some of the most compelling reasons. why you think that they should consider using Lumberyard on top of, you know, either the Unity or Unreal Engine, which, you know, seems to have a lot of momentum.
[00:05:10.789] Hao Chen: Yeah, so because VR is one of the new features that we added to Lumberyard, so we actually have a clean sheet of design when we decide to support VR. So what we ended up doing, and you'll see it in release 1.3, is a very clean integration of VR. So we basically abstracted the devices for you, So you only need to talk to our interfaces and not have to worry about the various device specific head tracking and input control because of that abstraction layer. Also we come up with this abstraction layer allow you to extend this to other devices that you might find a special need for and you don't really need us because the code is packed in a very compact way through a gem. Basically this is a concept for our engine for Lumberyard. which is a package engine module. So to support a new device, you just have to derive from our implementation, create a new gem, and you only pay for the feature that you need. So as a programmer, you know, I'm a programmer myself, I find our integration to be very lightweight, you know, so basically we don't add a whole lot of extra baggages you don't need, and it will give you all the essential functions wrapped around in a very tight package. The other thing is that because it's integrated into Lumberyard, so all of the Lumberyard goodies like Flowgraph, you know, visual scripting languages, automatically works. And so a lot of VR experiences using Lumberyard can be created without writing a single line of C++ code. You just enable the gem, make a few Flowgraph nodes, and it works with Oculus, it works with HTC Vive, and I think in the future, soon, you'll see that we'll support open VR as well. So even a lot more devices are supported. And we also have pretty good performance, although we are continuing to pay a lot of attention to improve the performance even further.
[00:07:06.493] Kent Bye: And is the Flowgraph visual scripting language, is that something that was already built into the CryEngine, or is that something that was specific to Amazon that you're adding in?
[00:07:14.430] Hao Chen: So CryEngine has a version of FlowGraph and we started with that version but we added a lot of features and rewrote a good portion of it to make it perform faster and also support the Lumberyard features better. And then now that's becoming one of the central thing that we gave access to developer without having to touch code or actually write scripting language. We also support Lua scripting as well, so it depends on how you want to use the engine. You can either write C++ code, or Lua scripting language, or use the FlowGraph visual scripting language. It's all up to you. So it started with the CryEngine, but like most features in Lumberyard, it starts with the CryEngine, then we improved it considerably, and we'll continue to improve it.
[00:07:59.263] Kent Bye: And can you talk about generally across the games industry how you've seen different games using Amazon Web Services in order to do either specific scaling out of multiplayer games and what some of those indicators were that led you to decide to start your own engine with Lumberyard?
[00:08:16.375] Hao Chen: So as a long-time developer of AAA games, the back-end has always been one of those very, very difficult problems for game developers, even for big developers. There are many very difficult issues. Scalability is a big one. Reliability is another one. And also you have all these capacities that you have to build out in your data center that are designed to accommodate the maximum throughput. For example, when Destiny launches, the day one, you reach a million concurrent users, and then it kind of goes down, right? So after it goes down, all that capacity that you build out in your data center is wasted. So cloud means many things to a game developer. It's cost-saving features because you're only paying for what you need, and it scales up and down. But it goes a lot more than that. It's also reliability. So Amazon Web Services is known for building these ironclad services that doesn't go down, or very, very rarely goes down. So it's a magnitude more reliable than any data centers out there. But also a lot of expertise that is required to make game backends. It takes a very large team of dedicated back-end engineers to create something that just works, and even more engineers to make it maintainable and scalable. One of our instincts is that we want to take all that know-how that we have for cloud, combine it with the knowledge of games. Most of the people in the games group are either cloud experts, cloud engineers, client engineers, So there is a mix of game people with a lot of knowledge about the game space and a lot of cloud knowledge. When we marry these two, what we're hoping and we are delivering a lot of solutions that are geared towards game developers to allow them to utilize the cloud to create things that they cannot create before, create games that are bigger than ever before, and create games that are very reliable. And all that through very few lines of code. You know, like a lot of our services, just a few lines of code. You know, Gamelift and Cloud Canvas are all working that way. So as a game developer, you know, I think it's a great thing. for the game community to be able to utilize the vast power of the cloud without having a huge large team of cloud engineers or networking engineers. And this also allows a lot of small and medium developers to get into the big games. You know, if you're a small group, then you used to only target very small casual games, maybe a very small online games and without knowing how to scale that up. And now it's like with Lumberyard, a small group can create a game that can scale up as big as you want it to. If you got the right idea and you got the right customer base, you just scale. And so all that is the power of Lumberyard and the cloud. And I think we're only seeing the tip of the iceberg here.
[00:11:13.452] Kent Bye: Yeah, I spent a couple years working at Puppet, which is a configuration management system, and so I'm aware of different automated ways of configuring cloud servers, either through Chef scripts or through Puppet declarative language or through Docker containers. And so I know that people who are running and automating, figuring their whole stack of the cloud, there's sometimes people who want to really dial into that level. When you talk about Lumberyard, is there a GUI interface to be able to do some of the lower level configuration management systems? Or maybe you could talk about the interface between Lumberyard and then what you're actually doing on the back end of the cloud.
[00:11:49.144] Hao Chen: There are several levels that you can use in Lumberyard and the cloud. So right off the back, if you just want to use the cloud services and configure them using a visual scripting language, we'll have Cloud Canvas. So this is sort of the highest level of using the cloud. It's actually very usable, you know, if you actually download Lumbre and try it, it's very intuitive, very usable, and you essentially have all the AWS services connected, you know, through this visual scripting language by dragging nodes around, right? And then underneath it, there's native APIs, you know, like AWS offers C++ API. So for the really hardcore guys who wants to use just the AWS primitives in the most direct way through their C++ language, that's available for you as well. Then there are also standalone services, which is not just code but also the console and the dispatch and the monitoring analytics, and that's Gamelift, which is, again, it's a package solution for you. You just have to provide your game binaries and it will automatically scale your back end for you. So that's another layer that you could utilize, which is this standalone services. And so that will become the model for us to add other cloud features, because if we want to Again, as a programmer, I completely understand what you are asking for. Sometimes you just want to get to the raw bits and say, I want to use this in the most direct way, and that's available. That will continue to be available. But for other people, this convenience is welcome. And it's not just a time-saving thing, it's an empowering thing. Empowering people who doesn't know how to use the cloud you know designer for example could script services like you have a social designer or a monetization designer they could tinker together web services using cloud canvas and then amazingly it works and so it empowers a lot of people who are now cloud expert to not have to worry about the underlying technology, just worry about what you're trying to create. And I think that's where I feel excited about what we're offering. But going back, yes, for sure, we will have low-level access to all of the cloud primitives.
[00:13:56.849] Kent Bye: I think in the early days of VRChat, they were running on a single server and it was somebody's birthday. So they started singing happy birthday to them and they had a situation where everyone was speaking at the same time and it crashed the server. And so you get this challenge with doing multiplayer social interactions online that At a certain point, the pipes kind of get clogged up in terms of just not having the capacity. And so I'm curious of some of the auto-scaling or features that you have within this Lumberyard integration to deal with many different people in a social situation, and if there's kind of a cap in terms of how many people you can be co-located in one location that may not be paralyzable.
[00:14:39.532] Hao Chen: Yeah, so with Gamelift, you could ask the server to dispatch as many servers as you'd like. And so it doesn't magically do everything for you. So you still have to decide, you know, how many capacity that you're going to run per instance. But when you do have that logic in there, then you say, OK, now I've reached my capacities. I want to spin up another fleet of servers. And that is painless for you. That's managed through our Gamelift console.
[00:15:05.955] Kent Bye: But that sounds like a manual process, though, that you have to see it breaking before you flip the trigger.
[00:15:10.886] Hao Chen: there is always going to be a bit of logic on the game side that you have to decide how your content maps to the cloud. So we don't try to make that decision for you, mostly because everybody's game is different. So if we started adopting an approach that's one size fits all, then you're going to find that it's neither efficient, nor it does what you really want. So we'd rather give you the ability to scale, but then have you maintain that logic. That being said, a bunch of other things that we're working that will hopefully one day allow you to give you even more choices to automatically scale your workload. Like if you have a very giant world of many, many people living in a persistent world, right now there's no solution for you to do that, to automatically handle that kind of traffic. So there are a lot of things that we're considering and working on. you know, that hopefully that one day will allow you to create new possibilities. But right now, going back to what Lumberyard currently offers, is that we have a very, very good way for you to scale your multiplayer backend, but you do have to kind of own the logic of where your game code, you know, maps to your backend.
[00:16:20.540] Kent Bye: And I know another trend that's happening within the virtual reality and gaming field is that there's going to be more and more artificial intelligence interactions, whether you're doing natural language processing or if you want to start to do offloading of deep learning or maybe sort of creative procedural content generation. using the cloud as a extra compute or GPU processes to facilitate additional artificial intelligence specific applications. And so, does Amazon offer any sort of natural language processing APIs or anything specific in terms of doing like specific deep learning AI type of functions to be integrating with Lumberyard?
[00:16:57.266] Hao Chen: Yeah, so Amazon does have Alexa, which is a platform. Alexa is a very powerful platform for voice recognition and understanding what you say and parsing what you say. Amazon also have a capacity through the AWS cloud of doing deep learning. So we have CPU and GPU racks. Both are prime for deep learning and we'll probably have in the company some of the world experts in deep learning as well. But in terms of a ready solution for game AI, utilizing deep learning, or procedural generations, or other type of things that utilize deep learning, we only started to get into those areas. It's a hot area, so I really can't talk about what we're working on, but I would just stop and say that it is one of the area that would definitely take notice, and there are work in that area that's continuing, but we're not ready to talk about them yet.
[00:17:52.763] Kent Bye: Yeah, I think that's probably one of the more compelling use cases for using the cloud, especially if you think about mobile and the limited resources that may be available for mobile. And so when you use Lumberyard, is there also any additional graphics processing that's happening in the cloud? Or is that something that you're kind of relying upon someone's local machine, whether it's a mobile headset or a desktop headset?
[00:18:14.374] Hao Chen: So our current mobile solution does not use the cloud as an accelerator, but it's one of the active areas of research. There are a lot of problems that we have to solve, latency is a big one, and also just figuring out what workload makes sense to run in the cloud. And we have lots of thoughts in there, and we have started to experiment in that space, but they're not ready for customer yet. But it is a good area of research, and it's definitely one of the areas we're thinking of utilizing the cloud is to bring up the quality of mobile rendering so that you can experience things similar to PCs and dedicated console. Yeah, so again, it's one of the areas that we're thinking, we're experimenting, working on, but we're not ready to talk about them yet.
[00:18:58.358] Kent Bye: Yeah, and just looking at the demo and knowing that it's still early days, and I'm just curious if you've seen any implementations or applications of people doing full virtual reality applications with Lumberyard so far in the beta program, or even internally.
[00:19:13.518] Hao Chen: So we have not announced some of the, we do have customers that wants to use Lumberyard, but we're not ready to announce them yet. And then you will hear that in the future when we're ready to announce. But I could say that yes, Lumberyard is being considered by a lot of customers to use because the cloud connection, because it's a full AAA engine, because it has full source, you know, which is a very powerful thing for people.
[00:19:38.844] Kent Bye: Yeah, so talk a bit more about the source code and what kind of license it's released under and what that enables people to do.
[00:19:46.157] Hao Chen: It's full source, so really there's no strings attached. No royalty, it's really free. A lot of people don't really grasp the idea, but it is free. There is, in the licensing, there is a requirement that if you use a public cloud, if you do intend to use a public cloud, it has to be AWS cloud. But other than that, you can treat it as your engine if you want to. And you have, as a programmer, you have full source code and can rewrite parts of it and add new features, suit your needs.
[00:20:14.366] Kent Bye: Great, so what's next for Lumberyard? What's some of the big next milestones that you have coming up here?
[00:20:19.128] Hao Chen: Yeah, so we want to have a constant trickle of features added to Lumberyard. And this is what we're doing release after release. Hopefully in every release you'll find something that we're doing to address some of the customer feedback. Whether it's that bug fix, whether it's the performance, whether it's adding new features, whether it's workflow improvements. So we want to kind of view Lumberyard not as a you release it one time and then a couple years later release another version but rather a continuous refinement that leading to greatness and so I really can't talk about like what is next big release because you know you can essentially view it as every installment is a release and we rather have that model so that we can respond to customer feedback you know sort of in a more prompt way. There's a lot of things that we're working on that are geared towards long-term future, a lot of R&D effort that's going on. Again, at Amazon we don't really talk about things that are not ready yet, and so you'll hear them in the future. But one thing I would say is that We are in this for a long haul so it's a big commitment in Amazon and we're here and we're here to stay for a long time with our customers and you're going to see continuing refinements of what we offer and more features that will come online.
[00:21:36.605] Kent Bye: Well, in terms of the target user, I know that VR is such a vast space, and I've been covering it for the last couple of years, and I have my own kind of way of thinking about the different verticals, but there's entertainment and medicine and enterprise uses, social applications, even virtual commerce and identity. travel, education, there's all these different domains that people are going to be using virtual reality in and I'm curious if you're like really focusing and starting on gaming or if you're trying to hit the full market of all the different applications for VR.
[00:22:07.361] Hao Chen: And so let's separate it into two things. One is Lumberyard's VR integration as a VR engine. So Lumberyard's customers are really everybody, right? So we want you to use Lumberyard in gaming and pretty much any place that uses VR and 3D. And so basically we want Lumberyard to be a very general-purpose engine. It is optimized for real-time games, and it's connected to the cloud for multiplayer games, but the same technology base can be used for anything. We do have customer feedback and asks that are not about games, and we're happy to support them, whether that's educational, architectural, or medicine, or whatever. So as Lumberyard, we want to provide a set of really clean, powerful VR features that allow you to do whatever that you want to do. Whether that's entertainment or not, it's up to you. Now, Amazon does have a vast interest in commerce and education and other areas. And VR is one of those platforms that potentially allow us to serve our customers really well in those areas. And so there are work and thinking in that space from Amazon as a company. So from both fronts, I think what you're saying is relevant. For Lumberyard, we want to support people doing non-games type of VR. We're focused on games, of course, but we support people, you know, to use Lumberjacks in creative ways. And if we have customers who have the scenario that we have not thought of, they come talk to us and we will, you know, see if we can accommodate them, you know. At GDC, there was this gentleman who came and asked us whether we could render diamonds in VR, because his business is selling jewelry. And so he wants really high-quality rendering of jewelry and diamonds in VR. And so a customer like that, even though they are small, we want to support them. And then, of course, VR as a platform for Amazon is one of the things that we're currently actively thinking about.
[00:24:04.608] Kent Bye: Is there an existing solution within the suite of different AWS offerings that would do like an e-commerce solution? Or is this something that you think that you would have to kind of do a specific Lumberyard integration in order to have people to be able to sell virtual goods?
[00:24:19.290] Hao Chen: Again, those are the stuff that we have thought about and experimenting, but we're not really ready to talk about how we're offering them yet. It's Amazon tradition to really think from the customer and backwards. So we think about what kind of things that they would like to do and what benefit they the ads and we go work backwards to work out technology that's required to support them. And because of that, we tend not to kind of speculate on things that we're working on until we have seen and convinced ourselves that it adds customer benefit. So in terms of answering your question directly, we don't have current Lumberyard offering that is directly tied to Amazon e-commerce platform. It is one of those things that we're considering.
[00:25:06.165] Kent Bye: Well, one of the other things that's come up here at SIGGRAPH is digital light fields and the computational resources that are required in order to actually render a light field and that. It seems like that's a perfect use case for using cloud computing in order to render light fields and, you know, I'm talking to Otoy as well as NVIDIA and looking at their solutions. It's just something that takes a lot of resources to either pre-render ahead of time of looking at it or, if you're doing it on the fly, you really need to have some sort of external cloud computing. Just curious if Lumberyard is something that you would integrate light fields within that engine, or is it something that's a completely different software stack in order to really deal with digital light fields and rendering them in real time?
[00:25:45.903] Hao Chen: Yeah, live fields are very interesting. We're very interested in live fields. As you noted, the live fields are difficult in both of capturing live fields, compressing of live fields, and then also delivering live fields and displaying it on real-time devices. All three of them are very difficult and some of them map well to the cloud compute. For example, storage and compressing and processing of live fields. And also delivering this compressed results to end devices is something that AWS is probably uniquely qualified to become the compute power and storage power behind it. But in terms of actually integrating live fields rendering into Lumberyard, it's one of those future research ideas that we're thinking about. But we'll talk more about it when we're ready.
[00:26:31.984] Kent Bye: Great. And finally, what do you see as the ultimate potential of virtual reality, and when it might be able to enable?
[00:26:40.790] Hao Chen: So, the way that I look at virtual reality is from different perspectives. From the end customer perspective, from the developer perspective, and also from engineering perspective. And if you look at the end customer perspectives, this idea that you're in the virtual space and presence, it's very refreshing. It's not new, but now that it's here, people started to take it for granted. But it enables the end consumer to have a lot better experiences. Closer to some of the things that you experience in real life, but perhaps even better than real life. Because in a lot of cases, real life is boring. And now we can give you better than real life and put you right there in that presence. From storytelling, from entertainment, from commerce, from education. Lots and lots of things just make intuitive sense. We are three-dimensional beings. We move and we react and perceive things in three-dimensionally much better than two-dimensional. So from end-user we see this is for customers of Amazon and Lumberyard, we see this as a big thing for them as a lot of interesting capabilities. from a content perspective this is a new canvas for creative people and so Lumberyard for one is what we're trying to give these pallet of tools to developers so that they can paint or create these worlds that are out of this world and so VR is just another canvas for them because now they can incorporate presence, they can incorporate scale, sense, you know like contacts that they could never before. So we think this is a very exciting area for the right tools and the content people can now visualize things and be able to create scenarios that are either very difficult for them to create before or even impossible. But we have only scratched the surface of what creative people can do with VR. And we just want to provide the right tools for them to let them express and have these amount of shackles around them. And then from an engineering perspective, this is pretty exciting too. It's new toys for engineers, new performance to be gained. A lot of things that VR is pushing us to do benefit other developers as well. Now that 60 hertz is like a minimum. It used to be the high bar, right? In the future we want to render 120hz or 240hz that allow us to think very deeply of how to do our engines the right way. How do we structure our engine and a lot of them are fundamentally going to be retargeted because we want to be high performance. And also in the future, you know, all these VR devices are likely going to be untethered, you know, like mobile-like devices. So we've got to think ahead of time how do we use cloud to make those devices render, you know, beautiful things. And I have to think ahead of time how to decouple all these frequencies of updates so that you can have things updating, you know, lighting update at 10 hertz and your shading happens at 60Hz and your input happens at 240Hz. And all that have to require some kind of rethinking. So as an engineer, I'm really excited about VR and AR because it pushes us to solve problems that didn't have to solve before. And so, you know, you see this boost of innovations coming in the next few years, which will benefit everybody. Even if you don't make VR games, you know, your performance is going to be greater and you have, you know, a lot more flexible. So ultimately, I think it's about our customers. We always come back to, what does VR do for you? If VR allows your kids to be educated better, VR allows old people to read better, and VR allows poor people to get entertained, and then we want to be there. So that's why we think VR is important.
[00:30:26.016] Kent Bye: Great. Is there anything else that's left unsaid at Lumberyard or anything about VR that you'd like to say?
[00:30:31.968] Hao Chen: I would encourage everyone to download Lumberyard and try our VR integration. If you have Oculus Rift or HTC and soon if you have any open VR headset, download it, make a level, throw a few flow graphs together and then put a headset on. It's really that easy. Enable a gem and put your headset on and you're in the VR world. We'd love to hear how you think about current VR integrations. If you're a programmer, we'd like to see how you think about our architecture of VR. If you're content people, we'd like to think about how our workflow is. And if you're an end consumer or enthusiast, make something fun. And one thing that really motivates us is to see people try things and use the engine in ways that we have not anticipated. And that's always motivated me in my career is to see content people and users come up with cool stuff. So my desire is to see as many people try Lumberyard today and try VR.
[00:31:31.373] Kent Bye: Awesome. Well, thank you so much. Thank you. So that was Hao Chen. He's the Senior Principal Engineer for the Amazon Lumberyard Project. So I have a number of different takeaways about this interview is that first of all I do think that multiplayer experiences and having social interactions within VR is really super compelling and I think that the biggest open question that I still have after doing this interview is what is the kind of pricing model going to be in order to actually sustain some of these types of experiences. So For example, if you look at Gamelift, which is this out-of-the-box multiplayer system that Amazon is providing, kind of direct integration with the Amazon Lumberyard game engine. So on their website, they have some kind of breakdown of costs. So it assumes that if you have like 5,000 daily active users, then it could be over $2,300 a month in the server fees. And so it's possible that in the short term that that could be cheaper for you than to do all that on your own. So it's likely that in the short term, it could be faster to get up and running and to be able to scale even higher and faster with using the cloud. But for some people, it may actually be cheaper for them to just set up their own hosting service and to roll your own. And it could be cheaper if you know what you're doing. If you don't know what you're doing, you'll have to either pay someone to do it, which is going to be really expensive, or use Amazon Web Services, which in the end could actually be cheaper. So there's a lot of different trade-offs, but the thing that comes to my mind is that if I had a multiplayer game and I'm charging people a fixed amount to be able to buy it initially, let's say I sell it to 5,000 people. Well, if those 5,000 people use it every month for like ever, then I'm actually going to at some point lose money. So it's like, how do you have some sort of way to pay to actually play? If doing multiplayer experiences is actually causing the developer to pay for that, then it could actually be that if the very active users are using it a lot, then what's the business model for people to actually sustain that? That to me is kind of like the question that I have coming away is that as a VR creators, as we're creating these social experiences, there's a cost to actually facilitating that. And, you know, Amazon Web Services is a way that's kind of like dead simple to start to get integrated, especially if I think you use the Amazon Lumberyard, there's going to be direct integration with a lot of these cloud backends. So maybe eventually all this will be kind of integrated in some sort of like low-level cryptocurrency so that as you play it's able to actually sustain the servers in the back end to be able to actually facilitate that. So I don't know, I think the business models for this type of thing You could start to see how it might make sense for a game developer, but when you start to think about the metaverse and what's it going to take to actually build and sustain the metaverse, then someone like Amazon would be certainly making a lot of money on actually making that happen. I suspect that there might be some sort of distributed web type of approach that could be taking the principles of a blockchain and being able to do more of a peer-to-peer type of approach to actually scaling out the metaverse. I think that's actually what High Fidelity is trying to do with some of their back-end ideas instead of kind of relying upon a centralized server system like the Amazon Web Service or maybe other solutions. So I think that's something that may emerge over time, but right now if you actually want to create a multiplayer experience then Doing something on a centralized server whether it's your own hosted server or something that's in the public cloud like with Amazon Web Services Certainly the thing that's gonna actually be available to be working today and so this was an interesting interview just because there is a number of things that I think that I can kind of extrapolate for what would be good things to do in the cloud like for example a multiplayer, which is already there and happening, but artificial intelligence I think is going to be a huge thing in terms of doing deep learning training, but also how I mentioned procedural generation of content, I think that's going to be something that's going to be tuned and cultivated within some sort of AI offering from Amazon, at least something that they're looking into and researching. And the voice recognition from Alexa I think is another use case in which if you want to have actually natural language interactions, then at this point there's a lot of the public cloud companies are offering different natural language inputs to be able to actually communicate. But again, the same question comes up in terms of how to actually financially sustain these types of applications and interactions. you may have like a fixed amount of people that buy it but if they use it with any significance then it's going to be a big question for how to actually sustain some of these programs in the long term. But some of the other things that I think would be interesting that I think that Amazon would be particularly well suited to handle is something like e-commerce solution to be able to handle some of the credit card transactions and you need to have some PCI compliant servers in the back end which takes a lot of effort and energy to set up and that's part of the Amazon Web Services offerings that they provide some of those back ends and so I could imagine that if you're doing some of these e-commerce transactions then they will likely be developing more integrations within the Amazon Lumberyard Also, I think the digital light fields are something that looks really quite amazing when you actually see them in VR. But like Hal said, there's kind of three different stages of capturing and compressing and delivering these digital light fields that are all really competitionally intense, which means that it's pretty well suited for the cloud. And so I'd imagine that Amazon's working on some specific solutions to be able to actually handle that, or at least that's on their agenda to be able to research. But again, how it's not saying that they're announcing anything just yet, but that to me would be something that would be kind of a logical fit for something that is using the strengths of cloud computing. And so Hal basically said that there's no one size fits all for being able to determine kind of the perfect formula for auto scaling your application, which essentially means that you're going to have to do some tuning to figure out if you're not reaching the capacity that you need, what are the triggers to be able to add more instances to be able to handle that capacity. And then once people go away, at what point do you kind of take away those extra instances? And so, So it doesn't sound like there's going to be a one-size-fits-all type of, like, automagically being able to autoscale. You're going to have to kind of figure that out on your own. So I did just take a quick look at some of the licensing for the Lumberyard. I did ask about it and Hal said that it was full source but it's not open source which essentially means that any modifications that you make you cannot distribute and so there's not going to be that extra dimension of kind of like this open source sharing of the code to be able to take other people's modifications and so Yes, while you may be able to modify it, you won't be able to share them back because that's not within the lumberyard terms of service and end user agreement. And so they are making it available, but you're not going to be able to actually share your own changes or take other people's changes or modifications. So I just wanted to clarify that, that it's still pretty much a proprietary license and not any free and open source license to it. And so that's all that I have for today. I just wanted to thank you for listening. And then if you enjoy the podcast, then tell your friends, spread the word. And if you'd like to help contribute to the podcast, then become a donor at patreon.com slash Voices of VR.