惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Attack and Defense Labs
Attack and Defense Labs
N
News and Events Feed by Topic
L
LINUX DO - 热门话题
PCI Perspectives
PCI Perspectives
www.infosecurity-magazine.com
www.infosecurity-magazine.com
爱范儿
爱范儿
D
DataBreaches.Net
Simon Willison's Weblog
Simon Willison's Weblog
S
Secure Thoughts
S
SegmentFault 最新的问题
博客园 - 【当耐特】
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
博客园 - 叶小钗
P
Proofpoint News Feed
The Hacker News
The Hacker News
T
ThreatConnect
N
News and Events Feed by Topic
T
Threatpost
The Register - Security
The Register - Security
WordPress大学
WordPress大学
博客园 - Franky
Recorded Future
Recorded Future
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
Project Zero
Project Zero
大猫的无限游戏
大猫的无限游戏
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
cs.CV updates on arXiv.org
cs.CV updates on arXiv.org
罗磊的独立博客
Stack Overflow Blog
Stack Overflow Blog
腾讯CDC
F
Future of Privacy Forum
F
Full Disclosure
Cyberwarzone
Cyberwarzone
J
Java Code Geeks
李成银的技术随笔
Schneier on Security
Schneier on Security
Know Your Adversary
Know Your Adversary
H
Hacker News: Front Page
人人都是产品经理
人人都是产品经理
博客园_首页
Scott Helme
Scott Helme
Google DeepMind News
Google DeepMind News
美团技术团队
Malwarebytes
Malwarebytes
Last Week in AI
Last Week in AI
T
Tailwind CSS Blog
T
The Exploit Database - CXSecurity.com
G
GRAHAM CLULEY
Recent Announcements
Recent Announcements
C
CXSECURITY Database RSS Feed - CXSecurity.com

CSS-Tricks

Revealing Text With CSS letter-spacing | CSS-Tricks Technical Writing in the AI Age | CSS-Tricks Cross-Document View Transitions: Scaling Across Hundreds of Elements | CSS-Tricks Cross-Document View Transitions: Scaling Across Hundreds of Elements | CSS-Tricks The State of CSS Centering in 2026 | CSS-Tricks Stack Overflow: When We Stop Asking | CSS-Tricks Cross-Document View Transitions: The Gotchas Nobody Mentions | CSS-Tricks What’s !important #11: 3D Voxel Scenes, Flying Focus, CSS Syntaxes, and More | CSS-Tricks Computing and Displaying Discounted Prices in CSS | CSS-Tricks rotateX() | CSS-Tricks rotateY() | CSS-Tricks rotateZ() | CSS-Tricks rotate() | CSS-Tricks Soon We Can Finally Banish JavaScript to the ShadowRealm | CSS-Tricks Using CSS corner-shape For Folded Corners | CSS-Tricks A Scrollytelling Gift for Mum on Mother’s Day 2026 | CSS-Tricks Google’s Prompt API | CSS-Tricks Making Zigzag CSS Layouts With a Grid + Transform Trick | CSS-Tricks Fixed-Height Cards: More Fragile Than They Look | CSS-Tricks What’s !important #10: HTML-in-Canvas, Hex Maps, E-ink Optimization, and More | CSS-Tricks The Importance of Native Randomness in CSS | CSS-Tricks contrast() | CSS-Tricks contrast-color() | CSS-Tricks Let’s Use the Nonexistent ::nth-letter Selector Now | CSS-Tricks Quick Hit #126 Recreating Apple’s Vision Pro Animation in CSS | CSS-Tricks Quick Hit #125 Enhancing Astro With a Markdown Component | CSS-Tricks Quick Hit #124 Markdown + Astro = ❤️ | CSS-Tricks Quick Hit #123 What’s !important #9: clip-path Jigsaws, View Transitions Toolkit, Name-only Containers, and More | CSS-Tricks A Well-Designed JavaScript Module System is Your First Architecture Decision | CSS-Tricks hypot() | CSS-Tricks The Radio State Machine | CSS-Tricks 7 View Transitions Recipes to Try | CSS-Tricks Quick Hit #122 Quick Hit #121 Selecting a Date Range in CSS | CSS-Tricks saturate() | CSS-Tricks justify-self | CSS-Tricks Quick Hit #120 Alternatives to the !important Keyword | CSS-Tricks Quick Hit #119 New CSS Multi-Column Layout Features in Chrome | CSS-Tricks Quick Hit #118 Making Complex CSS Shapes Using shape() | CSS-Tricks Quick Hit #117 Front-End Fools: Top 10 April Fools’ UI Pranks of All Time | CSS-Tricks Sniffing Out the CSS Olfactive API | CSS-Tricks What’s !important #8: Light/Dark Favicons, @mixin, object-view-box, and More | CSS-Tricks Quick Hit #116 Form Automation Tips for Happier User and Clients | CSS-Tricks Quick Hit #115 Generative UI Notes | CSS-Tricks Quick Hit #114 Quick Hit #113 Experimenting With Scroll-Driven corner-shape Animations | CSS-Tricks Quick Hit #112 JavaScript for Everyone: Destructuring | CSS-Tricks Quick Hit #111 Quick Hit #110 What’s !important #7: random(), Folded Corners, Anchored Container Queries, and More | CSS-Tricks 4 Reasons That Make Tailwind Great for Building Layouts | CSS-Tricks Quick Hit #109 Quick Hit #108 Abusing Customizable Selects | CSS-Tricks Quick Hit #107 The Value of z-index | CSS-Tricks Quick Hit #106 The Different Ways to Select <html> in CSS Quick Hit #105 Popover API or Dialog API: Which to Choose? Quick Hit #104 What’s !important #6: :heading, border-shape, Truncating Text From the Middle, and More Yet Another Way to Center an (Absolute) Element An Exploit ... in CSS?! Quick Hit #103 A Complete Guide to Bookmarklets Quick Hit #102 Loading Smarter: SVG vs. Raster Loaders in Modern Web Design Potentially Coming to a Browser :near() You Quick Hit #101 Distinguishing "Components" and "Utilities" in Tailwind Quick Hit #100 Spiral Scrollytelling in CSS With sibling-index() Interop 2026 Quick Hit #99 What’s !important #5: Lazy-loading iframes, Repeating corner-shape Backgrounds, and More Quick Hit #98 Making a Responsive Pyramidal Grid With Modern CSS Approximating contrast-color() With Other CSS Features Quick Hit #97 Trying to Make the Perfect Pie Chart in CSS Quick Hit #96 Quick Hit #95 CSS Bar Charts Using Modern Functions Quick Hit #94 No Hassle Visual Code Theming: Publishing an Extension Quick Hit #93
Lazy Loading Images in Svelte
CSS-Tricks · 2020-07-16 · via CSS-Tricks

One easy way to improve the speed of a website is to only download images only when they’re needed, which would be when they enter the viewport. This “lazy loading” technique has been around a while and there are lots of great tutorials on how to implement it.

But even with all the resources out there, implementing lazy loading can look different depending on the project you’re working in or the framework you’re using. In this article, I’ll use the Intersection Observer API alongside the onLoad event to lazy load images with the Svelte JavaScript framework.

Check out Tristram Tolliday’s introduction to Svelte if you’re new to the framework.

Let’s work with a real-life example

I put this approach together while testing the speed on a Svelte and Sapper application I work on, Shop Ireland. One of our goals is to make the thing as fast as we possible can. We hit a point where the homepage was taking a performance hit because the browser was downloading a bunch of images that weren’t even on the screen, so naturally, we turned to lazy loading them instead.

Svelte is already pretty darn fast because all of the code is compiled in advance. But once we tossed in lazy loading for images, things really started speeding up.

This is what we’re going to work on together. Feel free to grab the final code for this demo from GitHub and read along for an explanation of how it works.

This is where we’ll end up by the end:

Let’s quickly start up Svelte

You might already have a Svelte app you’d like to use, but if not, let’s start a new Svelte project and work on it locally. From the command line:

npx degit sveltejs/template my-svelte-project
cd my-svelte-project
npm install
npm run dev

You should now have a beginner app running on http://localhost:5000.

Adding the components folder

The initial Svelte demo has an App.svelte file but no components just yet. Let’s set up the components we need for this demo. There is no components  folder, so let’s create one in the src folder. Inside that folder, create an Image folder — this will hold our components for this demo.

We’re going to have our components do two things. First, they will check when an image enters the viewport. Then, when an image does enter, the components will wait until the image file has loaded before showing it.

The first component will be an <IntersectionObserver> that wraps around the second component, an <ImageLoader>. What I like about this setup is that it allows each component to be focused on doing one thing instead of trying to pack a bunch of operations in a single component.

Let’s start with the <IntersectionObserver> component.

Observing the intersection

Our first component is going to be a working implementation of the Intersection Observer API. The Intersection Observer is a pretty complex thing but the gist of it is that it watches a child element and informs us when it enters the bounding box of its parent. Hence images: they can be children of some parent element and we can get a heads up when they scroll into view.

While it’s definitely a great idea to get acquainted with the ins and outs of the Intersection Observer API — and Travis Almand has an excellent write-up of it — we’re going to make use of a handy Svelte component that Rich Harris put together for svelte.dev.

We’ll set this up  first before digging into what exactly it does. Create a new IntersectionObserver.svelte file and drop it into the src/components/Image folder. This is where we’ll define the component with the following code:

<script>
  import { onMount } from 'svelte';


  export let once = false;
  export let top = 0;
  export let bottom = 0;
  export let left = 0;
  export let right = 0;


  let intersecting = false;
  let container;


  onMount(() => {
    if (typeof IntersectionObserver !== 'undefined') {
      const rootMargin = `${bottom}px ${left}px ${top}px ${right}px`;


      const observer = new IntersectionObserver(entries => {
        intersecting = entries[0].isIntersecting;
        if (intersecting && once) {
          observer.unobserve(container);
        }
      }, {
        rootMargin
      });


      observer.observe(container);
      return () => observer.unobserve(container);
    }


    // The following is a fallback for older browsers
    function handler() {
      const bcr = container.getBoundingClientRect();


      intersecting = (
        (bcr.bottom + bottom) > 0 &&
        (bcr.right + right) > 0 &&
        (bcr.top - top) < window.innerHeight &&
        (bcr.left - left) < window.innerWidth
      );


      if (intersecting && once) {
        window.removeEventListener('scroll', handler);
      }
    }


    window.addEventListener('scroll', handler);
    return () => window.removeEventListener('scroll', handler);
  });
</script>


<style>
  div {
    width: 100%;
    height: 100%;
  }
</style>


<div bind:this={container}>
  <slot {intersecting}></slot>
</div>

We can use this component as a wrapper around other components, and it will determine for us whether the wrapped component is intersecting with the viewport.

If you’re familiar with the structure of Svelte components, you’ll see it follows a pattern that starts with scripts, goes into styles, then ends with markup. It sets some options that we can pass in, including a once property, along with numeric values for the top, right, bottom and left distances from the edge of the screen that define the point where the intersection begins.

We’ll ignore the distances but instead make use of the once property. This will ensure the images only load once, as they enter the viewport.

The main logic of the component is within the onMount section. This sets up our observer, which is used to check our element to determine if it’s “intersecting” with the visible area of the screen.

For older browsers it also attaches a scroll event to check whether the element is visible as we scroll, and then it’ll remove this listener if we’ve determined that it is viable and that once is true.

Loading the images

Let’s use our <IntersectionObserver> component to conditionally load images by wrapping it around an <ImageLoader> component. Again, this is the component that receives a notification from the <IntersectionOberserver> so it knows it’s time to load an image.

That means we’ll need a new component file in components/Image. Let’s call it ImageLoader.svelte. Here’s the code we want in it:

<script>
  export let src
  export let alt


  import IntersectionObserver from './IntersectionObserver.svelte'
  import Image from './Image.svelte'
  
</script>


<IntersectionObserver once={true} let:intersecting={intersecting}>
  {#if intersecting}
    <Image {alt} {src} />
  {/if}
</IntersectionObserver>

This component takes some image-related props — src and alt — that we will use to create the actual markup for an image. Notice that we’re importing two components in the scripts section, including the <IntersectionObserver> we just created and another one called <Image> that we haven’t created yet, but will get to in a moment.

The <IntersectionObserver> is put to work by acting as a wrapping around the soon-to-be-created <Image> component. Check out those properties on it.  We are setting once to true, so the image only loads the first time we see it.

Then we make use of Svelte’s slot props. What are those? Let’s cover that next.

Slotting property values

Wrapping component, like our <IntersectionObserver> are handy for passing props to the children it contains. Svelte gives us something called slot props to make that happen.

In our <IntersectionObserver> component you may have noticed this line:

<slot {intersecting}></slot>

This is passing the intersecting prop into whatever component we give it. In this case, our <ImageLoader> component receives the prop when it uses the wrapper. We access the prop using let:intersecting={intersecting} like so:

<IntersectionObserver once={true} let:intersecting={intersecting}>

We can then use the intersecting value to determine when it’s time to load an <Image> component. In this case, we’re using an if condition to check for when it’s go time:

<IntersectionObserver once={true} let:intersecting={intersecting}>
  {#if intersecting}
    <Image {alt} {src} />
  {/if}
</IntersectionObserver> 

If the intersection is happening, the <Image> is loaded and receives the alt and src props. You can learn a bit more about slot props in this Svelte tutorial.

We now have the code in place to show an <Image> component when it is scrolled onto the screen. Let’s finally get to building the component.

Showing images on load

Yep, you guessed it: let’s add an Image.svelte file to the components/Image folder for our <Image> component. This is the component that receives our alt and src props and sets them on an <img> element.

Here’s the component code:

<script>
  export let src
  export let alt


  import { onMount } from 'svelte'


  let loaded = false
  let thisImage


  onMount(() => {
    thisImage.onload = () => {
      loaded = true
    }
  }) 


</script>


<style>
  img {
    height: 200px;
    opacity: 0;
    transition: opacity 1200ms ease-out;
  }
  img.loaded {
    opacity: 1;
  }
</style>


<img {src} {alt} class:loaded bind:this={thisImage} />

Right off the bat, we’re receiving the alt and src props before defining two new variables: loaded to store whether the image has loaded or not, and thisImage to store a reference to the img DOM element itself.

We’re also using a helpful Svelte method called onMount. This gives us a way to call functions once a component has been rendered in the DOM. In this case, we’re set a callback for thisImage.onload. In plain English, that means it’s executed when the image has finished loading, and will set the loaded variable to a true value.

We’ll use CSS to reveal the image and fade it into view. Let’s give set an opacity: 0 on images so they are initially invisible, though technically on the page. Then, as they intersect the viewport and the <ImageLoader> grants permission to load the image, we’ll set the image to full opacity. We can make it a smooth transition by setting the transition property on image. The demo sets the transition time to 1200ms but you can speed it up or slow it down as needed.

That leads us to the very last line of the file, which is the markup for an <img> element.

<img {src} {alt} class:loaded bind:this={thisImage} />

This uses class:loaded to conditionally apply a .loaded class if the loaded variable is true. It also uses the bind:this method to associate this DOM element with the thisImage variable.

Native lazy loading

While support for native lazy loading in browsers is almost here, it’s not yet supported across all the current stable versions. We can still add support for it using a simple capability check.

In our ImageLoader.svelte file we can bring in the onMount function, and within it, check to see if our browser supports lazy loading.

import { onMount } from 'svelte'

let nativeLoading = false
// Determine whether to bypass our intersecting check
onMount(() => {
  if ('loading' in HTMLImageElement.prototype) {
    nativeLoading = true
  }
})

We then adjust our if condition to include this nativeLoading boolean.

{#if intersecting || nativeLoading}
  <Image {alt} {src} />
{/if}

Lastly, in Image.svelte, we tell our browser to use lazy loading by adding loading="lazy" to the <img> element.

<img {src} {alt} class:loaded bind:this={thisImage} loading="lazy" />

This lets modern and future browsers bypass our code and take care of the lazy loading natively.

Let’s hook it all up!

Alright, it’s time to actually use our component. Crack open the App.svelte file and drop in the following code to import our component and use it:

<script>
  import ImageLoader from './components/Image/ImageLoader.svelte';
</script>


<ImageLoader src="OUR_IMAGE_URL" alt="Our image"></ImageLoader>

Here’s the demo once again:

And remember that you’re welcome to download the complete code for this demo on GitHub. If you’d like to see this working on a production site, check out my Shop Ireland project. Lazy loading is used on the homepage, category pages and search pages to help speed things up. I hope you find it useful for your own Svelte projects!