Lenis smooth scroll

Lenis intercepts wheel input and eases the native scroll position. Sticky positioning, IntersectionObserver, and window.scrollY all keep working — which is why it's the right choice. The failures come from things that fight it, not from Lenis itself.

Setup

Vendor it rather than hot-linking a CDN, so the site stays self-contained and works from file://:

curl -sSL -o assets/vendor/lenis.min.js https://unpkg.com/[email protected]/dist/lenis.min.js

The dist/lenis.min.js build is an IIFE exposing globalThis.Lenis — fine in a plain <script>. (dist/lenis.mjs is ESM and will not work in a plain script tag.) Load it before your own script.

var reduced = matchMedia('(prefers-reduced-motion: reduce)').matches;
var lenis = null;

if (!reduced && window.Lenis) {
  var navH = parseInt(getComputedStyle(document.documentElement)
                        .getPropertyValue('--nav-h'), 10) || 0;
  lenis = new Lenis({
    duration: 1.05,
    easing: t => Math.min(1, 1.001 - Math.pow(2, -10 * t)),
    smoothWheel: true,
    syncTouch: false,        // leave touch native — smoothing it feels laggy on phones
    autoRaf: true,           // Lenis runs its own RAF loop; no manual ticker needed
    anchors: { offset: -(navH + 10) }   // in-page links clear the sticky header
  });
} else {
  // no Lenis — give the browser its own behaviour back
  document.documentElement.style.scrollBehavior = reduced ? 'auto' : 'smooth';
}

Required CSS

html { /* NO scroll-behavior: smooth — it fights Lenis */ }

html.lenis, html.lenis body { height: auto; }
.lenis:not(.lenis-autoToggle).lenis-stopped { overflow: clip; }
.lenis.lenis-smooth [data-lenis-prevent] { overscroll-behavior: contain; }
.lenis.lenis-smooth iframe { pointer-events: none; }

scroll-behavior: smooth on html must go. If you keep it, anchor jumps double-animate and fight the easing.

Trap 1 — `overflow: clip` kills `position: sticky`

This is the one that will cost you an hour.

lenis.stop() adds .lenis-stopped, whose CSS sets overflow: clip on <html>. That removes the scrollport, and position: sticky has nothing left to stick to — sticky elements snap back to their position in the document, i.e. the top of the page.

Open a menu at scroll 1200 and your sticky header renders at top: -1200. Off screen. Along with its close button. Users report it as "the close button disappeared."

The same thing happens with the classic body { position: fixed } scroll lock, for the same reason — no scrollport.

Fix: pin the header to the viewport for as long as scrolling is locked.

html.menu-open .nav-wrap { position: fixed; top: 0; left: 0; right: 0; }

document.documentElement.classList.toggle('menu-open', open);

Swap the class while the overlay is still opaque so the sticky→fixed layout change is invisible. Apply it on both the Lenis and the reduced-motion paths.

Test overlays at a non-zero scroll offset. At scroll 0 a broken sticky header is pixel-identical to a working one, so the bug hides completely. Verify at 0, ~1200, and deep in the page.

Trap 2 — don't stack two scroll locks

Pinning the body and calling lenis.stop() makes them fight: position: fixed on body collapses the page height, Lenis resyncs to 0, and closing the overlay dumps the user at the top of the page.

With Lenis, stop() plus its own overflow: clip holds the page without touching the body — so the scroll position is never lost and there is nothing to restore:

if (lenis) {
  open ? lenis.stop() : lenis.start();
} else if (open) {                       // reduced-motion path only
  scrollY = window.scrollY;
  document.body.style.position = 'fixed';
  document.body.style.top = -scrollY + 'px';
  document.body.style.width = '100%';
} else if (document.body.style.position === 'fixed') {
  document.body.style.position = '';
  document.body.style.top = '';
  document.body.style.width = '';
  window.scrollTo(0, scrollY);
}

Verify: scroll to 1200 → open → wheel → close. Position must read 1200 at every step, and scrolling must work again afterwards.

Trap 3 — anchor links

Lenis owns scrolling, so raw href="#id" jumps look wrong. Use the built-in anchors option with a negative offset equal to your sticky header height, or route clicks manually through lenis.scrollTo(target, { offset: -navH }).

Check the landing position: the section top must sit at or below the header's bottom edge, not underneath it.

Trap 4 — Lenis is usually not your jank

When scroll feels bad after adding Lenis, measure before blaming it. In one build native scroll actually measured worse (13% dropped frames vs 9%) — Lenis was fine and the real costs were elsewhere:

a fixed full-viewport mix-blend-mode overlay (film grain): 34% → 3% dropped frames just by removing the blend mode
backdrop-filter on a sticky element, which is on screen for every frame of every scroll
backdrop-filter on large elements generally — it re-samples per frame

Frame-timing harness

// in page
window.__f = []; let l = performance.now();
const t = n => { window.__f.push(n - l); l = n; requestAnimationFrame(t); };
requestAnimationFrame(t);

Drive it with real wheel events over the suspect section, then report mean, p95, and the percentage of frames > 33ms. Bisect by disabling one effect at a time with add_style_tag. Healthy is ~16.7ms mean and 0% long frames; anything over ~10% long frames is visible stutter.

brainzcode/premium-landing

premium-landing 是什么？

在你喜欢的 AI 中提问

文档