The case against the loading bar

Most of the time, a loading bar is the wrong artefact. It's a focal point that says nothing more interesting than "this is loading." If we can fill it with content instead, we should.

Our default pattern: the heaviest scene paints a low-cost first frame within 200ms (a SVG fallback or a static poster), then upgrades to WebGL when the texture, geometry, and shaders are ready. The user sees the composition immediately. The fidelity arrives without ceremony.

When the wait is unavoidable — large download, slow API, AI generation — we tell the user what we're doing in plain words, not what percentage we're done. "Loading the Lhotse summit" beats 47%. Time is the same; the felt experience is different.

The only place a percentage genuinely helps: very long uploads, where the user is going to walk away and come back. Even then, it should be calm.