JHDD Web Design Report — 2026.09.11

The first Three.js Conference in Paris signals a new phase for web-based 3D, moving from niche application to foundational craft.

Recent developments across the web front-end reveal a concerted evolution towards performant, real-time computational graphics. These articles are not isolated explorations of visual effects; they collectively trace a pattern where direct GPU access, sophisticated shader languages like TSL, and robust libraries like Three.js and MediaPipe are converging to enable deeply integrated, interactive experiences that push beyond traditional DOM constraints. This is about establishing new foundations for visual systems on the web.

The project Still, which creates a living generative garden using WebGPU and procedural systems, exemplifies this shift. Mainstream industry opinion often categorizes high-fidelity 3D on the web as either a performance burden or a luxury feature reserved for high-budget marketing microsites. This perspective is increasingly outdated. Still demonstrates how projects leveraging WebGPU and carefully architected procedural generation can be inherently performant, rendering complex scenes without relying on large asset downloads or inefficient DOM manipulations. The craft now lies in understanding how to harness the GPU directly for core interactive experiences, not merely for decorative flourishes.

The advancements detailed in “Building an Infinite Liquid Glass Grid” and “Drawing With Light: An Exploration of Lit GPU Tubes” further illustrate this deeper integration. The use of TSL for custom shaders grants web developers unprecedented control over rendering pipelines, enabling intricate effects that respond dynamically without the overhead associated with pre-rendered animations or complex JavaScript-driven DOM updates. Within two years, WebGPU will emerge as the standard rendering backend for a significant portion of interactive web components, subtly displacing many current canvas 2D or even basic WebGL 1.0 implementations. This will empower web professionals to design layouts and motion graphics with a level of fidelity and performance previously unattainable in standard browser environments.

The primary force resisting this integration is the entrenched reliance on DOM-centric workflows and CSS for all visual effects, coupled with an often-justified fear of performance degradation from unoptimized 3D. The perceived steep learning curve for GPU programming and shader languages also slows adoption among professionals comfortable with established front-end paradigms.

A working web design professional should dedicate time this week to dissecting a simple WebGPU example that utilizes TSL. The goal is not immediate mastery of shader programming, but rather to gain an intuitive understanding of how vertex and fragment pipelines process visual data. This foundational knowledge provides a critical new lens for approaching layout, motion design, and performance optimization, even when working primarily with HTML and CSS, by informing decisions about where and how to best leverage hardware acceleration.

---

Curated References