How to Bring 3D on the Web: from Games to Gaussian Splats
In this episode, Will Eastcott, the founder of PlayCanvas, takes us through the journey of making 3D on the web more accessible and shares practical tips on how to optimize textures and meshes for your project.
The newsletter is also shared with our 10k audience on LinkedIn and if you would like to become a sponsor just answer this email or contact me on LinkedIn 👉 Gabriele Romagnoli
Interview with Will Eastcott
What inspired you to create PlayCanvas, and what was the original vision for the platform?
Will Eastcott: Originally, PlayCanvas was all about indie game development. I wanted to create a tool that would make 3D game creation more accessible for smaller teams and individual developers. Over time, though, we realized that interactive 3D wasn’t just for games; there were so many other applications. Today, the platform is used for everything from advertising to product visualizers, yet gaming is still in our DNA and that foundational love for gaming drives a lot of what we do.
Why do you think web-based gaming hasn’t grown as fast as some might have wanted?
Will Eastcott: The big barrier has been monetization. On the web, it’s harder to implement effective revenue models compared to mobile apps or console games. You don’t have the streamlined in-app purchasing systems like iOS has, and it’s challenging to get people to pay for digital items or subscriptions online. Ads are an option, but they don’t work for every game. Without these kinds of financial incentives, it’s been difficult for web gaming to really explode, even though the technology is there.
What would you say are the main benefits of developing a game for the web?
Will Eastcott: Freedom is a huge factor. On the web, you’re not limited by app store rules or approval processes; you can launch when you want and iterate quickly. And the web’s audience is massive. Any device with a browser can access your game. If you nail down your SEO and get the right marketing going, you have a global audience ready to play instantly. It’s incredibly powerful, though building that expertise in areas like SEO and analytics is key for success.
How has the scope of PlayCanvas evolved beyond gaming?
Will Eastcott: PlayCanvas is now what I’d call a genre-agnostic engine. It started with games, but now we support all kinds of interactive 3D content. Automotive brands use it for product visualizers, companies create 3D configurators, and advertisers are using it for immersive web-based experiences. The demand for photorealistic rendering in web applications has pushed us to adopt cutting-edge graphics techniques and explore Gaussian splatting. We’re constantly evolving to keep up with these new use cases, while still catering to our gaming roots.
What types of developers and creators are currently using PlayCanvas?
Will Eastcott: It’s for a broad range of creators, from game developers who need advanced tools for things like animation and physics to brand marketers looking to create visually impressive configurators. Our editor is user-friendly and visually oriented, so it’s great for developers who like a hands-on, visual experience. It’s probably closer to a tool like Unity in that way, whereas more code-centric frameworks like Babylon.js or Three.js cater more to developers who are comfortable working primarily in code.
What optimizations do you consider essential for making 3D content work smoothly on the web?
Will Eastcott: Load time is crucial on the web. People will bounce if they wait too long, so we’ve worked hard on compression and asynchronous streaming in PlayCanvas. Techniques like Draco compression for meshes and Basis compression for textures are huge for reducing load times. On top of that, focusing on runtime performance to minimize CPU and GPU load helps keep devices cool and saves battery, which is vital for mobile users. Optimizing from the start is key—don’t wait until the end of the project.
Do you have any specific advice for optimizing textures and meshes for the web?
Will Eastcott: For meshes, it’s actually less about reducing vertices and more about managing how you’re filling pixels. You want to be mindful of factors like lighting, shadows, and post-processing effects, as these can really tax the GPU. For textures, use Basis compression whenever possible—it reduces memory use significantly and keeps texture data in a format that’s easy for the GPU to process. It’s about finding that balance between performance and visual quality.
Over the last year you have been experimenting with Gaussian Splatting. How does that fit into PlayCanvas?
Will Eastcott: Gaussian Splatting is a process of capturing reality in 3D by generating tiny particles, or “splats,” that represent an object’s shape and surface. These splats adapt to lighting and camera angles, creating a photorealistic effect. We saw this technology and immediately wanted to integrate it because it offers a new level of realism for web-based 3D content. With PlayCanvas, you can use Gaussian Splatting for interactive environments or even avatars, which is really exciting for our users.
What’s your current setup for capturing Gaussian splats?
Will Eastcott: For quick captures, I use apps like Polycam or Luma AI on my phone; they guide you through the process and are very accessible. For higher quality, I’ll use a DSLR camera, take around a hundred photos, and run them through the Inria tools, which are great for photogrammetry. But to do this at a professional level, you really need a good PC with an NVIDIA GPU to process the data efficiently. The tech has made it so that anyone can capture splats, but there’s room to refine the process for advanced results.
Once you have the splats, how do you process and clean them up?
Will Eastcott: That’s where SuperSplat comes in. It’s a tool we’ve developed to import, clean, and optimize splats. You can select and delete stray particles, transform splats in 3D space, and export them in various formats depending on your engine. It’s web-based and open-source, so it’s easy to use with PlayCanvas, Unity, or Unreal. Super Splat has really streamlined the cleanup process and made it accessible to a wider range of developers and creators.
For developers and creatives interested in trying out Gaussian Splatting, what’s a good first step?
Will Eastcott: Start simple—use an app like Polycam or Luma AI to capture a basic object. Then bring it into Super Splat for cleanup and optimization. Once you’re comfortable, try experimenting with higher-end equipment or more detailed environments. The whole process is becoming more accessible, and even with just a smartphone, you can capture something really impressive. From there, it’s about getting creative and exploring the tech’s potential in XR or other interactive 3D applications
Check out the full interview on your favourite platform
That’s it for today, and don’t forget to subscribe to the newsletter if you find this interesting
See you next week