For years, Unity was the default. Client wants an AR experience? Unity. Interactive 3D product viewer? Unity. Branded game for a campaign? Unity. It was the safe answer. And for a long time, it was the right one.
But safe answers age. And somewhere around my fifteenth AR campaign, I started noticing a pattern. Most of the time we spent was not on the experience itself. It was on the pipeline around it. Building for iOS and Android separately. Waiting for app store approvals. Convincing users to download a 200MB app to try a 30-second filter. The friction was killing the conversion rates our clients cared about.
So I made the switch. Not all at once, but project by project. PlayCanvas for the editor and rendering. WebXR for device access. TensorFlow.js for the ML-powered tracking. Everything running in the browser.
The first thing that changed was deployment. Instead of a two-week app store review cycle, I had a URL. Scan a QR code, tap a link, you are in the experience. No install. No storage permission. No update prompts. For marketing campaigns with tight deadlines, this alone justified the switch.
The second thing was performance, and this surprised me. PlayCanvas compiles to highly optimized WebGL. Our AR face filter ran at a locked 60fps on mid-range Android phones. The same experience in Unity, exported via WebGL, was hitting 30fps on the same devices. Part of this is engine overhead. Unity's WebGL export carries a lot of runtime weight that a purpose-built web engine does not.
The tradeoffs are real though. Unity's physics engine is better. Its particle system is more mature. If you are building a physics-heavy VR game, Unity is still the right tool. And the C# ecosystem gives you things like strong ECS patterns and battle-tested networking libraries that the web ecosystem is still building.
But for the work I do most - branded AR, product visualization, interactive web experiences - the web stack wins on every metric that matters to the business. Faster deployment, broader reach, lower user friction, and comparable visual quality.
The tooling gap is closing fast. PlayCanvas just shipped their new shader editor. Three.js has first-class WebGPU support. The browser APIs for camera, gyroscope, and haptics are stable and well-documented. I am not saying Unity is dead. I am saying that for a growing category of interactive experiences, you no longer need it.
If you are on the fence, start with one project. Pick something small, non-critical, and time-boxed. Build it with a web-native stack and compare. Not just the technical output, but the entire pipeline: build time, deployment, user reach, iteration speed. The numbers will speak for themselves.