The default answer for 3D on the web, and the substrate under most of the case files in this issue. Its role is to make WebGL addressable: scenes, cameras, materials and lights, rather than raw buffers and draw calls.

What separates award-tier usage from ordinary usage is almost never Three.js itself — it is what gets written into the materials. Active Theory's work runs full WebGL environments with custom GLSL; Lusion's Oryzo renders a single object with easing tuned to imitate physical inertia. Same library, entirely different ceiling.

The animation engine, and by consensus the standard for tying scroll position to anything — including 3D camera movement. If a site sequences reveals as you scroll, ScrollTrigger is the likeliest mechanism.

GSAP runs its own awards nomination for sites built with it, which is a reasonable proxy for motion quality specifically; Mat Voyce's kinetic-type portfolio was a 2025 nominee.

A smooth-scroll library, and the single cheapest upgrade in this stack. It supplies the weighted, slightly inertial scrolling that reads as expensive — the difference between a site that feels engineered and one that feels assembled.

Its authors are a working studio that also wins days on their own client work, which is a rare arrangement: the tool is maintained by people who ship award-tier sites with it. Use it deliberately, though — hijacking native scroll carries real accessibility and performance costs, and a jury testing usability will feel a badly tuned implementation immediately.

Shaders are where the top tier actually separates from the field. Custom GLSL is what produces materials, lighting and effects that no library ships by default, and it is the most common thread through Developer Award winners.

TSL is the significant recent development: shaders written once in it compile to both WebGPU and WebGL backends. That removes the reason studios were postponing WebGPU — nobody wanted to maintain two shader codebases for one site.

The successor graphics API, with a substantially higher performance ceiling than WebGL — compute shaders, and particle counts that WebGL cannot sustain. Adoption has been gated not on capability but on fallback strategy.

Utsubo's IVRESS, FWA Site of the Month for May 2026, is the working proof: WebGPURenderer with a WebGL fallback, one TSL shader codebase serving both. Expect the visual weight of award-winning 3D to increase noticeably as this pattern spreads.

The most under-used material in the stack. Immersive Garden's Cartier build carries a score as a narrative layer, not as ambience; Resn treats audio as a first-class design decision across their work.

Sound is disproportionately effective at producing the sense of a place, and disproportionately rare because it requires an interaction to unmute. Sites that solve that gate elegantly get an atmosphere their competitors cannot match visually.

Worth stating plainly because it undercuts a lot of the above: scroll-triggered reveals no longer require a library. Native CSS scroll-driven animations handle entry reveals, progress indicators and pinning with no JavaScript, no bundle cost, and automatic reduced-motion behaviour.

This will not replace ScrollTrigger for 3D camera sequencing. It does replace it for the eighty percent of sites whose scroll motion is fading and sliding elements into view — which is most of them.

The structural shift of 2026: real-time 3D is being documented as a Webflow technique, not solely as a studio capability, and Three.js now has a dedicated conference. What was a specialism practised by perhaps thirty studios worldwide is becoming a skill with a training path.

The consequence for the awards is predictable. As the technique commoditises, the juries move the bar — which is precisely why restraint-first sites like By-Kin are sweeping awards in the same cycle as heavy WebGL work.