Scenario and AWS on AI Models & Working With Game Developers

Introduction

80.lv: Please introduce yourself. How did you get into the industry? What have you been doing before Scenario?

Marie Gerard, Head of Sales at Scenario: I'm Marie Jarrett. I'm the Head of Sales at Scenario. I've been with Scenario for about officially a year and a half, unofficially I'd say two years. Basically managing all the sales initiatives but also the growth initiatives. 

Previously, before Scenario, I was working for Amazon Web Services as an Account Manager, managing very large amounts of customers. So, really helped those customers as well in their scalability challenges, making sure that they had the right resources for the right needs.

Before joining AWS, I co-founded a start-up, which was an art gallery in augmented reality back in London because I come from a fine arts background, specializing in Digital Arts. So, I was very much involved in everything that was like virtual reality and augmented reality, and thinking about its different applications, especially when it comes to visual arts.

The AI Models and Clients

80.lv: Please tell us more about the models you use for image generation. Do you just rent the infrastructure or do you rent out the models? Where do you get the material to train your models?

Marie Gerard: Scenario is basically the first-gen AI engine for the gaming industry. We enable gaming studios to train their own custom AI models based on their own datasets. These datasets can be visuals. Once they have trained these AI models, they can use text or images as references to generate an infinite number of visuals aligned with their artistic direction. This includes new props, characters, landscapes, backgrounds, and anything else they may need for ideation, production, or marketing assets.

80.lv: Can you tell us more about your clients? What types of companies are using your services?

Marie Gerard: To train an AI model, you actually need no more than 100 images, and in fact, most of the time you only need 20 images. So you don't need a very vast dataset. The AI is basically optimized as well to just take very few references that are inputted from those studios. We are talking all the way from indie developers to AAA studios. I would say generative AI can be used wherever people want it in their own workflows. So if you're an indie dev, you might have a need for assets because you don't have a huge artistic team to basically produce those assets, and so Scenario can be there to do that. And if you're a AAA studio as well, you might look at innovative ways to basically accelerate this creation process as well within your artistic teams.

Scenario & AWS

80.lv: Where does AWS stand? What are you using it for?

Marie Gerard: AWS is our Cloud infrastructure because obviously, you know, Scenario can run from anywhere and from any kind of device because we are using the cloud to do so. And basically, when you are looking at the production of assets as well, you would want it to be scalable to generate those images as well in the least amount of time possible. And you potentially need, like, hundreds of thousands to millions of those assets as well, depending on if you want to plug it directly into the games.

So I would say AWS today is helping us to really serve our customers at scale, but also in a secure way because we're benefiting from the AWS cloud and the AWS security grades as well. We are implementing all the security best practices that need to be put in place to serve indie developers, but more importantly, those AAA studios that are very particular and looking at the security and the way we handle their data.

Scenario is built as an API-first platform. Every single feature can be taken apart and integrated into any kind of platform. We have a web application, we also have an iOS mobile application, and an Android application as well. Or you can just take the API as it is, and some of our customers are doing that. You can directly integrate it within the games to produce those assets and deliver hyper-personalized content potentially to the gamers.

Going 3D

80.lv: Currently, you focus on the production of 2D images, are you planning to produce 3D assets as well?

Marie Gerard: Currently, it's the production of 2D images we focus on. 3D assets are definitely on our roadmap. It's a big challenge as well because there are so many different ways to consider 3D. Every game studio has different specs and requirements. You have different types of polygons, meshes, and textures. We are building solutions that would serve most of those use cases. Our workflow is focused on training your own AI model to produce 2D assets and then from those 2D assets, produce 3D assets as well.

Thoughts on AI

80.lv: What do you think about AI in general? How do you think it's changing the landscape? What is going to happen next?

Marie Gerard: That's a great question that I really love. So on a personal level, my motto was always to basically help technology empower artists. How can this be done? I think today, generative AI is going to become a commodity. We're not going to talk about generative AI anymore. It's going to become an entirely part of the artistic process. If you look back 100 years ago, or even if you look at the industrial revolution as well, when the first cameras started to pop up, artists were very reluctant at adopting cameras as a recognized art form. People were like, 'Hey, this is mechanical. This is reproducing my art and devaluing it.' And time has shown that the way images are spreading so fast is actually bringing more value to the arts itself.

So, I think that AI is going to become a commodity. It's going to become part of the workflow as well. I think artists are going to adopt it the same way that the camera today has fully integrated into the process of a painter. I don't know any painter today that is not using a photograph as a reference and projecting it onto a canvas. I think it's going to become exactly the same with generative AI. It's always going to be a tool for artists to really create even more.

Amazon Web Services

80.lv: Could you introduce yourself to our readers and tell us a little bit about the work that you do?

Dan Carpenter, Director of the North American Games Segment at Amazon Web Services: I've been with AWS for about 12 years, and I currently lead the AWS for Games sales organization. Our team works with studios of all sizes – from startups to AAA studios – to bring them the best solutions, and combinations of AWS products and offerings from our partners, that enable the industry to create the next generation of interactive experiences.

AWS is comprised of over 200 fully featured services, which include games-focused services like Amazon GameLift. Developers would rather be building fun, innovative games, versus spending time and effort handling infrastructure. We see customers run a variety of workloads with AWS, including global game servers, storage of game code and assets, content distribution, development pipelines, and unique AI/ML applications. We help studios build games faster, better understand their player community, and efficiently manage their games as they grow. Many of the world's largest game studios, including but not limited to Epic Games, Roblox, Riot Games, EA, Nintendo, Rovio, and FromSoftware, use AWS today, and we're really proud to continue to expand how we work with different studios across the industry.

AWS for Games

80.lv: Tell us a little bit about the AWS solutions for games. What does it provide? What is this product at its core?

Dan Carpenter: We think about AWS for Games solutions within three pillars: building a game, running a game, and growing a game. For building a game, AWS offers solutions such as virtual workstations, version control, build pipelines, and graphics rendering with third-party partners that help developers create assets faster. When running a game, there are very large-scale events that take place where developers may not know how many people will be playing the game. Being able to rapidly scale servers to meet player demand, manage costs, and reduce latency, is very important during the initial launch of a title. AWS provides scalable game server solutions that can be deployed in any of AWS’s 99 Availability Zones within 31 geographic regions. This helps studios bring games closer to players, reduce latency, and enhance the overall player experience. To help studios grow games, we work closely with them to develop analytics pipelines and understand what their live operations look like. This could help them understand how players interact with their game or where players might get stuck in a level.

Across our services, security is paramount. We focus on perimeter security, content distribution, protection of assets, and allowing both security-encrypted storage solutions along with user access controls that can help studios build games across geographies and not worry about who has access to underlying files. Also, importantly, we help studios to protect players against toxicity through our Community Health solutions. AWS content moderation services and solutions provide automation and AI capabilities to implement a reliable content moderation solution without requiring machine learning expertise. For game studios, these tools can help prevent hate speech, profanity, bullying, and other behaviors that lower safety and engagement in gaming spaces.

Artificial Intelligence and AWS

80.lv: Can you tell us a little bit about the way AWS can help improve and help AI development?

Dan Carpenter: Amazon has invested heavily in the development and deployment of AI and machine learning (ML) for over two decades for both customer-facing services and internal operations, from the recommendation engines that personalize the shopping experience on Amazon.com to the AI-powered robots that optimize order fulfillment in our warehouses. We believe that generative AI is going to be transformative to game development and the cloud is going to unlock it by providing services to game developers of every size, at scale.

AWS has a variety of services we make available to game studios, whether for advanced use cases, such as deploying GPU-based instances to build and train their own large language models, or simple applications like screening imagery in real-time, monitoring text and chat, or translating player interactions across multiple languages. AWS provides fully-managed services like Amazon SageMaker that can abstract many foundational elements to build a model.

AWS also recently announced Amazon Bedrock, which enables game studios to take pre-trained foundation models from AI startups like AI21 Labs, Anthropic, and Stability AI, or Amazon Titan foundation models, and fine-tune them privately and securely using only a few examples. Using Amazon Bedrock, none of the customer’s data is used to train the original base models, and since all data is encrypted and does not leave a customer’s Virtual Private Cloud (VPC), they can trust that their data will remain private, and confidential.

Additionally, we are already seeing the value of generative AI supporting code creation via Amazon CodeWhisperer, which supports 15 leading languages, many crucial to game development, including C, C++, C#, Go, Ruby, Rust, Python, Java, and JavaScript.

Working With Game Developers

80.lv: Can you share a little about the way AWS works with game developers? How do you structure fees, how do you help developers onboard?

Dan Carpenter: AWS has teams of dedicated professionals that can engage with studios to answer questions about how AWS works, train staff how to use AWS solutions, or go in and sit side-by-side with developers to help them construct applications from the ground up. We can help developers scale out their infrastructure and then work with them from a strategy standpoint on what the long-term growth of a game should look like. AWS also has a huge ecosystem of industry partners from a solutions vendor perspective, including Incredibuild, Perforce, AccelByte, Adobe, and Epic Games’ Unreal Engine. These are solutions that developers are likely familiar with, and they’re available on AWS. We also have third-party consulting organizations that can augment an existing development team with resources, and help them construct these solutions.

In addition to that, we have an entire education arm that can provide and facilitate classroom-style instruction, or virtual training for those that want to learn more about the AWS ecosystem at their own pace. We have over 200 different services natively that are available to developers, and our staff can help them construct what these different types of application layers look like and how to leverage those solutions in an all-encompassing solution. AWS has pledged to train 29 million people around the world with free cloud training by 2025, and we have introduced games-specific training via a cloud game development learning plan. Once someone starts running with AWS, we provide a dedicated team of account executives, solution architects, and product specialists to align directly with those customers. We want to make sure our customers understand how to leverage AWS from the get-go and address multiple different applications of our service in real-time as they need. One of our key tenets is that we leave no developer behind.

The Advantages

80.lv: What do you feel is the biggest advantage of partnering with AWS?

Dan Carpenter: Amazon has multiple decades of experience in the games industry and is dedicated to this space. You can see this through not only AWS, but Amazon Games, Prime Gaming, Amazon Luna, and more. Many of the world’s biggest game studios use AWS, in part because it provides the broadest set of functionalities, one of the largest communities of partners, and consulting support to help them advance their journey and scalability. No other infrastructure provider offers all of the global locations that put resources closer to players and ensure the most optimal gaming experience overall.

We’re also one of the most secure clouds. We are audited by compliance programs globally, and we can also help studios navigate what it looks like to deploy a game anywhere in the world. Additionally, AWS enables a faster pace of innovation as we continue to advance our generative AI offerings. Finally, studios can rely on us to innovate on their behalf. 90% of what we’re building comes directly from the feedback of our customers, with the other 10% focused on what we believe will benefit them the most in the future. We look forward to continuing to help game developers build the future of gameplay faster and deliver low-latency, responsive gameplay experiences to millions of players around the world, on any gaming device.

Marie Gerard, Head of Sales at Scenario

Dan Carpenter, Director of the North American Games Segment at Amazon Web Services

Interview conducted by Kirill Tokarev