Ben Houston's Blog

Ben is currently available for contract work for 3D & web solutions — reach out.

This is the personal blog of Ben Houston (bio here), a 3D computer graphics software developer. For a quick overview of the work I am proud of, see my portfolio. I've worked on projects like Web3DSurvey, Three.js, Deadline, Krakatoa, and Exocortex VFX plugins. I'm also an entrepreneur, having founded Land of Assets, Threekit, Clara.io, Exocortex, and Frantic Films Software.

Latest writing

November 11, 2026

Pluggable Indirect Lighting for Three.js

Part of the Quest for AAA Quality in Three.js Series

Games do not agree on one way to compute bounced light and reflections. They pick a stack that fits the game. This is a proposal for two configurable slots in Three.js's WebGPURenderer: one for indirect diffuse, one for indirect specular, each an ordered list of sources that fall through by coverage.

Featured Posts

Transferable Pixel-Perfect Material Graphs for Blender & Three.js
Launching Land of Assets
Web 3D Survey v2
Zstdify: Pure TypeScript Zstd

November 10, 2026
Introducing Houston Stratified Noise
Blue-noise textures make stochastic screen-space effects look good, but they cost megabytes to download or tens of seconds to generate. Houston Stratified Noise fills the same role with integer arithmetic in the shader: zero bytes, zero startup.
November 10, 2026
Jittered Sample Weights
A small buffer, lots of samples, no banding: the usual deal is that you can't have all three. Jittering the sample weights lets you have your cake and eat it too.
November 10, 2026
Adding Robust Light Baking to Three.js
Part of the Quest for AAA Quality in Three.js Series
Three.js could always display a lightmap, but not make one. Here is the baking pipeline now in Three.js: a real-time UV unwrapper, a lightmap baker, a ray-traced radiosity solver and a denoiser, plus the glTF extensions and HDR formats needed to store and ship the result.
November 9, 2026
Building the AAA Lighting Pipeline in Three.js
Part of the Quest for AAA Quality in Three.js Series
From the G-buffer to temporal upscaling, here is each stage of the lightmap, probe and screen-space reflection pipeline measured against Three.js: what games require of it, what Three.js has today, and the pull requests in review or still to write.
November 8, 2026
How AAA Games Light Indoor Scenes
Part of the Quest for AAA Quality in Three.js Series
DOOM, Call of Duty, Frostbite and Killzone light a room in remarkably similar ways: baked bounce light, probes for moving objects, and a cascade of reflection sources that fall through by coverage. This is a study of what shipped, and of the pipeline a web renderer can take from it.
November 7, 2026
AAA Quality on a Web Budget
Part of the Quest for AAA Quality in Three.js Series
A AAA game targets a console and ships tens of gigabytes. Three.js runs from one code base on a mid-range phone and ships with a web page. Before choosing techniques, here are the constraints: which APIs, the memory usage, and the memory bandwidth required for a 60 fps frame.
November 5, 2026
Introducing ies-tools: Making IES Light Files Easy to Use
An IES file captures the shape of a real light, but opening it gives you a wall of numbers. I built ies-tools to make seeing that light as easy as opening an image.
November 5, 2026
Coding Agents Underestimate the Cost of Code They Own
Your coding agent just rebuilt a library you could have installed. Now every future change has to pay for that decision.
November 4, 2026
Bump Maps Need Units
Move the camera closer and the bumps get flatter: Three.js can change a material’s apparent depth without changing the material. The missing convention is simple—height needs a unit and a space.
October 30, 2026
Don't Fear Your Competitors
A polished competitor isn't proof of a strong business. Lessons from Deadline, Clara.io, and Threekit on validating demand, assessing real threats, and pricing on value.
October 28, 2026
GPU Stratified Annealing: An Experiment in AI-Guided Blue Noise Optimization
My AI-guided search for better blue noise began with a spectacular result that failed broader tests. The eventual sampler reduced filtered error by 10.7% in a Three.js reflection scene—but lost to FAST on rectangular-light shadows.