— Journal · RSS
Notes from the studio.
Writing on technology, design, VR, AI — and what it takes to ship them. Mostly long, sometimes short, occasionally wrong.
Journal · · 7 min
Gaussian splatting in 2026: ready for looks, not for measurement
Gaussian splatting produces the most photorealistic 3D captures available — reflections, soft detail, real light. But in 2026 it is production-ready for some jobs and not others. Here is where it wins over Matterport and photogrammetry, and the four limits that still decide the call. Journal · · 8 minWhat is a digital twin? A plain guide (and what one costs to build)
A digital twin is a living virtual replica of a real place or object, kept in sync with reality. Here is what that actually means, how we build one from a 3D scan, the real use cases, and what it costs in 2026. Journal · · 7 minVirtual tours are not just marketing: onboarding, training and the TPF case
Most virtual tours are sold as marketing. The more useful ones never face the public at all — they onboard new staff, run evacuation drills, and let partners walk a facility from another country. The operational case for 3D capture, and what we built for TPF. Journal · · 8 minVirtual Tours That Convert: A Practical Playbook
Do virtual tours increase conversions? Sometimes — when the space is the product and the embed is built to be found, fast, and accessible. Here's our honest playbook on when a 3D tour earns its place and when it's just expensive decoration. Journal · · 7 minMatterport vs Gaussian Splatting vs Photogrammetry in 2026: how we choose
Matterport, Gaussian splatting and photogrammetry solve different problems. Our 2026 decision rule for which 3D capture method to use, with real tradeoffs. Journal · · 8 minHow to 3D Scan a Heritage Building: Our Field-to-Web Workflow
How to 3D scan a heritage building: a step-by-step workflow covering site survey, lighting, capture pattern, scale references, hard surfaces, processing and web delivery. Journal · · 5 minWhy we still scan with Matterport in 2026
Photogrammetry got cheaper. Gaussian splats got everywhere. We kept reaching for the same Pro2 we've used since 2019 — here's the unromantic reason.— The journal, by email