The default answer for 3D on the web, and the layer underneath most of the case files. It lets a designer work in the vocabulary of a film set — scenes, cameras, materials, lights — instead of issuing drawing instructions to the graphics card one at a time.

What separates award-tier usage from ordinary usage is almost never Three.js itself — it is what gets written into the materials. Active Theory builds whole environments with surfaces written from scratch; Lusion's Oryzo turns a single object, weighted so it moves as though it had real mass. Same library, entirely different level of result.

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 best work separates from the merely competent. A shader is a short program that decides what every pixel of a surface looks like — how it catches light, how it distorts, how it moves. Writing your own is what produces materials and effects no off-the-shelf library includes, and it is what no amount of assembling parts will get you.

The recent development that matters: those surface programs can now be written once and translated automatically to whichever engine the visitor’s browser is running. That removes what made studios postpone the new engine — having to write and maintain everything twice for one site.

The replacement for the engine that has drawn browser 3D for a decade, and a far faster one — it can hold scenes with orders of magnitude more moving pieces than the old engine manages. What held it back was never what it could do; it was what to serve to the visitors whose browsers cannot run it.

Utsubo's IVRESS, FWA Site of the Month for May 2026, is the working proof: the new engine where the browser has it, the old one where it does not, and a single set of surface programs 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.

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 raise their expectations — which is precisely why restraint-first sites keep winning in the same cycle as heavy WebGL work.