GitHub - elder-plinius/GL4SS: the looking glass — spatiotemporal image + video engine
Short summary
The Looking Glass is a browser-based tool that uses AI to generate images and short films of any spot on Earth at any selected year (from the Great Dying to 3050) and hour. It offers a dial of 284 time stations, a movable sundial for light conditions, and a WebGL “wormhole” animation between eras. All processing runs client-side with local storage, no backend or account required.
- It generates AI images/video of any location on Earth at any time from 66 million years ago to the year 3050, controlled via a 284-station dial.
- Everything runs client-side with no backend or account—you supply your own OpenRouter key and data is stored locally in IndexedDB.
- It uses a WebGL wormhole shader for transitions between eras and lets you compare two time periods via swipe or blink-compare.
- It's open-source (AGPL-3.0-or-later) and includes curated presets like Pliny sailing toward Vesuvius or Apollo 11 liftoff.
Full summary
The Looking Glass is a browser-only spatiotemporal viewer that can reconstruct any spot on Earth at any hour from the Great Dying (66 mya) up to the year 3050. You drop a pin on a map, spin a dial through 284 “stations” (big jumps in deep time, one-year steps for recent history), drag a sun-and-moon control around the rim, then pull a physical-style lever to generate a still or short film with sound. No server calls beyond your own OpenRouter key, no account, no hidden backend—your frames live in IndexedDB, light up on the dial when you revisit them, and stay free unless you decide to prerequest the next station.
That single-billet interface combines Leaflet maps and world imagery, a fixed-lamp optic for unified lighting, and WebGL for a wormhole-style transition shader whose direction and color shift with time travel. You can pin two eras side by side with a swipe seam or blink-compare by holding space. Presets let you jump straight to scenes like Pliny sailing toward Vesuvius in AD 79, a green Sahara, drowned Alexandria, Apollo 11 liftoff, even a moonlit Moai quarry. Four runtime libraries—React, React DOM, Leaflet, three.js—plus five image-generation and five video-generation models selected in settings, all governed by a strict CSP.
Error messages pinpoint the problem in plain English—wrong key, out of credits, models retired—and only offer fixes that will work. Controls rely on clicks and keyboard shortcuts (←→ to step stations, J for journeys, M for place, P or space for compare). It’s open-source under AGPL-3.0-or-later, built to give you total control over your own temporal archive without telemetry or surprise charges.