Minimal CSS-only blurry image placeholders | Hacker News Reader

Minimal CSS-only blurry image placeholders

leanrada.com

470 pointsChiptuneIsCool75 comments

This is really cool, I love seeing folks use CSS in clever ways. :)

My one feedback would be to avoid using attr selectors on the style attribute like [style*="--lqip:"]. Browsers normally lazy compute that string version of the style attribute [1], but if you use a selector like this then on every style recalc it'll force all new inline styles (ex. element.style.foo = bar) to compute the string version.

Instead if you use a separate boolean attribute (or even faster a class) it'll avoid that perf foot gun. So write <div lqip style="--lqip: ..."> and match on that.

[1] https://source.chromium.org/chromium/chromium/src/+/main:thi...

see also the author's last note on the upcoming parsing feature of `attr()`, which would solve both problems (performance and verbosity) at once:

    <img src="…" lqip="192900">

I can't wait for the day when attr has this implemented. It would eliminate the need for so many inline styles

Is this really a performance concern though?

It's obviously mostly an aesthetic nitpick for this blog post specifically and not the project itself, because few people are going to be exploring the encoded space outside of the blog post, but the sliders letting you explore the LQIP space would "flash" a lot less if the base color was encoded in the high bits instead of the low bits.

Since there're independent Lightness values set for each section (I'd say quadrant but there are 6 of them), I wonder if two bits can be shaved from the `L` value from the base color. It'd take some reshuffling and might not play well with color customization in mainly flat images, but I think it could work.

I'm also curious to see that they're doing solely grayscale radial gradients over the base color instead of tweaking the base color's `L` value and using that as the radial gradient's center, I'd imagine you'd be doing more math that way in the OKLab colorspace which might give prettier results(?).

Tempted to play around with this myself, it's a really creative idea with a lot of potential. Maybe even try moving the centers (picking from a list of pre-defined options with the two bits stolen from the base color's L channel), to account for varying patterns (person portraits, quadrant-based compositions, etc).

Was expecting the common "background-image: data url + filter: blur" that a lot of static site generators produce, not a binary encoding algorithm implemented in CSS! Very impressive.

I wonder what other things could be encoded this way. Those generic profile pictures, perhaps? (The ones where your email or account id is hashed to produce some unique geometric pattern.)

Really like this, nice work!

Something to note is that Color Theif (Quantize) is using median cut on RGB, it would be interesting to try and extract dominant color in OKLab instead.

I also love the idea of a genetic algorithm to find an ideal match for a given image; it should be possible to simulate radial gradients server & client side with webgpu, but probably overkill for such a simple task.

EDIT: Although it works for me in Chrome, it doesn't seem to work in Safari v16.1.

Forgive my ignorance, feel like it's embarrassing to ask here to be honest, but can someone explain how this helps/works? I've never actually used these placeholders, but I always imagined that they work by processing the image beforehand on the server and using something like a super low quality image or gradient or such as the placeholder. If this is done in pure CSS, does the browser not need to download the image first to figure out what's in it, before then doing the placeholder effect? Perhaps it doesn't help that I've not had my morning coffee yet, but I don't understand.

It’s still computed at build time or dynamically, by a programming language. The “pure CSS” part of it means that the hash is decoded into something visual by CSS without any JavaScript required.

These placeholders are generated by processing the image on a server beforehand. Generally they create some html, css or svg markup that is served inline. Having to do a separate request for such placeholder is very bad idea.

It's not clear if these placeholders do actually help, especially placeholders with very low quality. In my opinion, they only add visual noise.

I'd focus more on avoiding layout shifts when images load, and serving images in a good format (avif, webp) and size (use `srcset` or `<picture>`).

> It's not clear if these placeholders do actually help

Well, it depends what you mean by help. It’s very dependent on use case and desired UX. Obviously you can prevent layout shifts without them, you can provide feedback on loading status in other ways, and ensure images don’t slow load time without colored placeholders. But they can provide a pleasant UX for some use cases, when done right. They can be annoying when not done well.

There are certainly legitimate cases where the placeholders come from a separate request. Most CRDTs and similar sync engines you don’t want to (and/or are not allowed to) store binary images directly inside them, so you need to store references to some other blob storage. But Blurhash is a simple short string (and LQIP here is a simple integer) and those store well in CRDTs and other sync engines, so you can pair that with your reference pointer (which might not even be a URL depending on your blob storage engine and its sync mechanics and authorization schemes and whatever else) and whatever other metadata you want/need to include like width/height or aspect ratio and alt/title/caption.

When the CRDT or document sync engine inevitably sync much faster than your blobs you have something to show in that placeholder. If the blob sync fails for some reason, you still have something to show more interesting than your browser’s old broken image logo under your “Sync is slow or broken” warning.

I think placeholders help a bunch in situations like that where your image fetch is a lot more complicated than a URL that you can add in a `src` attribute. It’s also really easy to get into situations where such blob fetching is complex: In cases where you have to respect user and/or tenant privacy and need complex OAuth flows. In cases where you need end-to-end photo encryption. In cases where you need peer-to-peer sync and only P2P sync because you’ve been mandated to reduce touch points and likelihood of accidentally storing photos at rest in middle layers. Situations like images of HIPAA data, PII, PIFI, etc.

On a static site with public (or cookie sessioned) images direct linked by URL, yeah the placeholders don’t do much other than check certain design boxes. There’s lots of other places images (and their metadata) come from, and placeholders are a useful fallback in the worst cases.

Here's the server-side (Node.js) build script that calculates the integer placeholder image values and adds them to the document: https://github.com/Kalabasa/leanrada.com/blob/7b6739c7c30c66...

Very nice solution!

Definitely very low resolution, but compared to sites that use a solid color this seems much better. And only requiring one variable is really nice.

The article seems very well thought through. Though for both the algorithm and the benchmark algorithm the half blue / half green image with the lake shows the limitations of this technique. Still pretty good considering how light weight it is.

The half blue / half green image still looks better with LQIP than BlurHash. I was getting ready to use BlurHash in my app, might try this instead!

In fact, LQIP looks better than most of the BlurHash examples in the gallery (https://leanrada.com/notes/css-only-lqip/gallery/); not sure if these were cherry picked or what.

Author here: Definitely cherry picked ;)

I did deliberately pick some "bad" examples like the blue+green image, and other multicolor images.

I wanted to add an upload function so people could test any image, then i realised I'd have to implement the compression/hashing in the client. Maybe i should!

I tried getting that working earlier using Claude to convert your script - you can see the result here: https://claude.site/artifacts/b747d94a-2923-4904-8ed1-7330bf...

Here's the transcript and code: https://claude.ai/share/4a562082-b681-4f0c-909c-3c32c34fd050

I could tell and I really appreciate it. It's really helpful to see both the good and the bad.

Great work!

Just in case anyone also misses it like I did, dark reader (at least on firefox) appears to apply itself to the final colors causing them to look quite bad and not match the input image at all. I would have discounted this entirely if it wasn't for all the praise I was seeing in the comments here.

Cool hack, but performance is terrible. That page makes scrolling on my phone laggy.

We need to embrace WebP v2 for this kind of stuff. I took one of their images, resized it to 24x16px, and compressed it with Squoosh at 65% quality. It compresses to just 144 bytes. And it looks way way way better than these CSS gradients.

https://squoosh.app/

Cool app but no maintained library to use it in our own apps and scripts.

https://github.com/GoogleChromeLabs/squoosh

If you go to NPM the first thing it says in bold is "no longer maintained" and it links to the Google Chrome labs GitHub as well.

https://www.npmjs.com/package/@squoosh/cli#project-no-longer...

So to clarify, is the GitHub maintained but the npm distribution is not? Or is none of it maintained unless you use the website/app?

Another simple css-only solution as the article mentions is gradients. Like

  background: linear-gradient(
    to right,
    #51463e 0%,
    #28241f 100%
  );

Tool: https://tools.w3clubs.com/gip/

cool, but the fact that you can now do this with CSS is part of the reason that a new browser engine is so unlikely - one of 100000 things that css can do now and need to be supported :(

Maybe we should have kept CSS simple and JS optional. Maybe we took a few wrong turns...

It's all additive so each new feature does indeed add complexity, but my impression is that it's often the older features and all their quirks which are the most difficult to implement. Adding a few math functions is much easier than ensuring compatibility with CSS2 floats.

Years ago before you could do anything this fancy with CSS I experimented with generating 3x2 pixel images server side and then presenting them as base64 encoded pngs in a "scoped" block of CSS to ensure they loaded before the src images. Coincidentally this was the same 3x2 layout as OP did here with CSS. I abandoned it because a 3x2 image scaled up looked terrible, and went with average color instead. This solution looks a lot better visually.

I still do the average color thing today since it's easy to calculate and store server side (I resize the uploaded image to 1x1 px and just record the result as a hex code in the DB).

This works significantly better than I would have expected. I was just exploring extremely simple png strings as an alternative to the hash libraries requiring decoding. I had also explored two color css gradient, based on pregenerated major/minor colors, but too course to be useful (for a fast scrolling gallery). I’ll give this a test drive!

Why is the page so sluggish on mobile?

Probably because of all of the wildly complex CSS calculations it's running, as described by the article.

Yep, there are a lot of layers and compositing operations (maybe more than necessary?). I suppose it could be simplified further.

Unusably slow for me on my Librem 5. I'm talking about two or three fps.

I read the title as “Minimal CSS — only blurry image placeholders” first by mistake. ;)

broken on iOS (iPad)

Worked for me in Mobile Safari in iOS on my iPhone.

Maybe it’s iOS version dependant. I’m a bit out of date (on purpose for jailbreak) and the demo is broken for me.

I'm up to date and it's broken for me :/

Same setup, didn't work. (Empty space where blur supposed to be.)

Also broken for me:

Safari 18.0 (20619.1.26.31.6), macOS Sequoia 15.0

Me too. Thumbnails just appear black.

Safari 17.6 (19618.3.11.11.5), MacOS Sonoma 14.7.3 (23H417)

It works on Chrome on the same machine.

Works fine on macOS Sequoia 15.4 with Safari 18.4.

It's a cool solution, and I like that it's CSS only. But the generated placeholders are way too blurry/lossy for my personal preferences.

My attempt at the four color approach:

https://0x0.st/820Q.html

Man, it's wild how much you can do with CSS calculations. How long before someone makes a CSS-only Game Boy emulator?

I did find this Idle Game entirely in CSS amusing: https://lyra.horse/css-clicker/

Love the website and article! Looks like even with CSS there's always new things to learn and do, good stuff.

No one remembers the lowsrc img attribute.

Amazing work! Thanks you for sharing!

This is brilliant!

That's sexy!

this is super neat! love your site

I suppose the existence of bad uses does not invalidate the good but it feels like 99% of blurry image placeholder behavior is actually just preventing people from seeing anything unless they also run the ad and spying javascript that monetizes the site.

So a CSS-only way is neat and indisputably better but I think it's missing the point? The point of blurry placeholders isn't to make things easier or display better. The point is to make things worse. This write up is definitely making things better.

The point of blurry placeholders is to support loading a page with potentially hundreds of images (maybe with additional lazy loading, which doesn't need JavaScript these days) without blocking display of the page on loading those full images.

I'm not sure why you think it has anything to do with forcing people to execute JavaScript?

Can't speak for everyone ofc, but not sure if I ever wanted blurry placeholders when images load fast enough, or found them anything but annoying when not. I think these bells and whistles only serve as designer's self-affirmation.

I've definitely wanted them on photo galleries with large numbers of thumbnails, and I appreciate them when they are implemented well, especially if I'm on a slow connection.

Agreed, they just create needless visual activity. How about a page specify where the images appear, and leave it up to the browser to decide how to show them and load them? Is that too simple and workable?

Indeed. So the visitor need only wait for the 20mb javascript bundle, but not the 600kb of images, before he can see the 1kb of text that he visited the site to read.

Sounds like you're in favor of a version of blurry placeholders that's implemented in less than 1KB of CSS.

The placeholder is inline in the markup. It can be displayed before the image loads. Which is not inline. I have no idea what 99% thing you're talking about.

Look at literally any "newspaper" website. From the smallest local paper to the NYT.

That's a whole separate thing with a different reason for existing.

It is really simple to make sure your blurred css placeholder is cosmetic, and a progressive enhancement. I would know, I wrote one a month ago for my personal site.

My goal was to have something that'd transmit all the essential bits of the site in the first 14kB, and worked basically on everything. It wasn't hard, honestly.

It wasn't a particularly complex site but i guess what I'm saying is any well done (and well intentioned) implementation of blurred image placeholders will works with or without JS. That is just sound engineering...

This is pretty neat

..or use tailwind - https://tailwindcss.com/docs/filter-blur

That's not at all equivalent to what the OP is doing. The point isn't just to blur an image, which is what those Tailwind classes do, the point is to render a very compact blurry version of an image which hasn't loaded yet.

I know very little of css and to me it seems like a configuration file for rendering text, similar to changing default fonts ornsizes for matplotlib plots using plt.rcParams. How does this do inage blurring then?

If you read the article, it explains exactly how the technique works.

One of the many ways CSS allows you to customize formatting is to change the background style of elements. In addition to just using a solid color or image, you can specify a procedural gradient. And by superimposing several such gradients, you can make a very blurry approximation of an image.

CSS also includes a basic expression language which allows evaluating simple arithmetic expressions. So you can encode all the blurred image's parameters as a packed integer in a single compact CSS property per image, and use rules to define the gradients in terms of that integer.

Note that CSS is not used to compute the blurred image representation itself -- you have to do that separately. (Even if you could do it in pure CSS, the whole point is to show a blurred preview image before the image itself is downloaded to the browser, so doing it in CSS would defeat the purpose.)

> [It] seems like a configuration file for rendering text...

A more accurate mental model might be, "a declarative language for styling HTML elements," where "styling" is very broad. You can make user interfaces that show and hide elements, have animations, etc. triggered by clicking buttons without a single line of JavaScript. It's a lot more powerful than the configuration parameters to plotting functions, in my book it's a programming language rather than a configuration language.

Nice, but... it's not actually minimal. But nice.

Also a bit of nitpicking. While it provides a visual placeholder for an image that's being fetched, it does not reflect its content. So, when it's loaded we can see a completely different color palette and shapes.

What do you mean? In my opinion this library does a very good job of representing the image’s color palette considering it’s encoded into a single integer. (Even smaller than usual because of CSS limitations, only 20 bits!)

You don’t even need JavaScript to decode that integer into the image. The underlying CSS may be complex, but for the user of the library it definitely feels minimal in a good way.

In the gallery https://leanrada.com/notes/css-only-lqip/gallery/ there's a good example of what I mean - the bottom right image or 4th from the end. A completely different image in comparison to the gradient.

As for the minimalism, I understand what you mean. But I understood the "minimal" part in regard to implementation, not usage. If we only mean usage, we can say the same about a lot of libs, that they are minimal. Yeah, it's minimal for the end user, but under the hood it is not as minimal. It's not anything bad, it's just how I interpreted the title.

That's pretty much an issue with every LQIP solutions though, including BlurHash and ThumbHash. The only thing that matters is that it's close enough to the real thing, since they're meant to serve as placeholders.

I see two issues, let's say "opportunities":

First is the limitation to one hue value. Something like the Sunflower (blue + yellow) is just yellow. Maybe there's a tradeoff that could pack more hue but with less luminescence.

The second is how the primary color is selected. Several images (plant on grey background, street food vendor) appear to be averaging across the image and getting a grey value. By selecting better for the predominant color and its placement, the greys would appear on their own.

Reply on news.ycombinator.com