AAlicia lu·Nov 2Houdini Copernicus to GLB: Nine Bakes and 18 Browser StartsNine Copernicus bakes and 18 browser starts test normal maps, GLB repairs and PNG versus KTX2 payloads on an RTX 5070 Ti.11 min read·digitalgarden-alicia.hashnode.dev
AAlicia lu·Nov 1Three.js r187 compileAsync: 40 Desktop Starts, Measured40 Chrome starts compare r186 and r187 compileAsync: WebGPU timing, long tasks, WebGL2 fallback and hidden-object limits. Raw data included.11 min read·digitalgarden-alicia.hashnode.dev
AAlicia lu·Oct 31Spark Radix vs Spark 2.3.1: What a 5M-Splat Desktop Test ShowsSpark Radix and Spark 2.3.1 tested at 1080p on an RTX 5070 Ti: 54 trials, WebGL2 sort delay, real draw calls, format failures and migration limits.11 min read·digitalgarden-alicia.hashnode.dev
CCoding Macaw Bootcamp·Oct 30Federated Learning in 2026: Architecting Privacy First AI with Next.js and RustLearn to build privacy first Federated Learning system using Next.js, WebGPU, Rust. Master distributed machine learning for FAANG engineering roles.4 min read·codingmacawbootcamp.hashnode.dev
AAvinash Kori·Oct 29Building an in-browser AI background remover: licences, WebGPU limits and a 17-buffer surpriseTransformers.js, BiRefNet on WebGPU, an ormbg CPU fallback, and why your photo never needs to leave the browser6 min read·avinashkori.hashnode.dev22CG
AAlicia lu·Oct 27Three.js WebGPU Grass Wind: One Field, 54 Desktop Trials54 desktop trials of shared TSL wind, 100k–1M grass blades, Houdini wind masks and chunk recycling in Three.js WebGPU and WebGL2.13 min read·digitalgarden-alicia.hashnode.dev
RRun Cheng·Oct 26Simulating N-Body Galactic Collisions and NASA TESS Exoplanet Transits in the Browser with WebGPU & WebGL2When I launched Space Simulator Online as a zero-install 3D Solar System and night-sky planetarium in raw WebGL2, the most frequent request from physics teachers and students was: "Can we go beyond st4 min read·runcheng.hashnode.dev
AAlicia lu·Oct 24Three.js GLB Loading: Cache, Decode and the First GPU FrameTwenty Three.js WebGPU trials compare raw and meshopt GLBs, separate cache, decoding and compilation, and retain the first-run outlier.5 min read·digitalgarden-alicia.hashnode.dev
YYunus Emre Vurgun·Oct 15Agents That Live in the Tab: A Practical Guide to Browser-Side AI Agents with Local InferenceHow to build browser-side AI agents with local inference via WebGPU and WebLLM: architecture, tool loops, privacy and practical code.15 min read·yunusemrevurgun.hashnode.dev
ZZyVOP·Oct 3MicroLLM: Let Your Users Bring Their Own AI KeysMicroLLM lets users bring their own AI keys so developers can avoid token bills. We also look at browser-based LLMs as a key-free alternative.15 min read·blog.zyvop.com
