Image Fragmentation Effect With CSS Masks and Custom Properties

Avatar of Temani Afif
Temani Afif on

Geoff shared this idea of a checkerboard where the tiles disappear one-by-one to reveal an image. In it, an element has a background image, then a CSS Grid layout holds the “tiles” that go from a filled background color to transparent, revealing the image. A light touch of SCSS staggers the animation.

I have a similar idea, but with a different approach. Instead of revealing the image, let’s start with it fully revealed, then let it disappear one tile at a time, as if it’s floating away in tiny fragments.

Here’s a working demo of the result. No JavaScript handling, no SVG trickery. Only a single <img> and some SCSS magic.

Cool, right? Sure, but here’s the rub. You’re going to have to view this in Chrome, Edge or Opera because those are the only browsers with support for @property at the moment and that’s a key component to this idea. We won’t let that stop us because this is a great opportunity to get our hands wet with cool CSS features, like masks and animating linear gradients with the help of @property.

Masking things

Masking is sometimes hard to conceptualize and often gets confused with clipping. The bottom line: masks are images. When an image is applied as mask to an element, any transparent parts of the image allow us see right through the element. Any opaque parts will make the element fully visible.

Masks work the same way as opacity, but on different portions of the same element. That’s different from clipping, which is a path where everything outside the path is simply hidden. The advantages of masking is that we can have as many mask layers as we want on the same element — similar to how we can chain multiple images on background-image.

And since masks are images, we get to use CSS gradients to make them. Let’s take an easy example to better understand the trick.

img {
  mask:
    linear-gradient(rgba(0,0,0,0.8) 0 0) left,  /* 1 */
    linear-gradient(rgba(0,0,0,0.5) 0 0) right; /* 2 */
  mask-size: 50% 100%;
  mask-repeat: no-repeat;
}

Here, we’re defining two mask layers on an image. They are both a solid color but the alpha transparency values are different. The above syntax may look strange but it’s a simplified way of writing linear-gradient(rgba(0,0,0,0.8), rgba(0,0,0,0.8)).

It’s worth noting that the color we use is irrelevant since the default mask-mode is alpha. The alpha value is the only relevant thing. Our gradient can be linear-gradient(rgba(X,Y,Z,0.8) 0 0) where X, Y and Z are random values.

Each mask layer is equal to 50% 100% (or half width and full height of the image). One mask covers the left and the other covers the right. At the end, we have two non-overlapping masks covering the whole area of the image and, as we discussed earlier, each one has a differently defined alpha transparency value.

We’re looking at two mask layers created with two linear gradients. The first gradient, left, has an alpha value of 0.8. The second gradient, right, has an alpha value of 0.5. The first gradient is more opaque meaning more of the image shows through. The second gradient is more transparent meaning more of the of background shows through.

Animating linear gradients

What we want to do is apply an animation to the linear gradient alpha values of our mask to create a transparency animation. Later on, we’ll make these into asynchronous animations that will create the fragmentation effect.

Animating gradients is something we’ve been unable to do in CSS. That is, until we got limited support for @property. Jhey Tompkins did a deep dive into the awesome animating powers of @property, demonstrating how it can be used to transition gradients. Again, you’ll want to view this in Chrome or another Blink-powered browser:

In short, @property lets us create custom CSS properties where we’re able to define the syntax by specifying a type. Let’s create two properties, --c-0 and--c-1 , that take a number with an initial value of 1.

@property --c-0 {
   syntax: "<number>";
   initial-value: 1;
   inherits: false;
}
@property --c-1 {
   syntax: "<number>";
   initial-value: 1;
   inherits: false;
}

Those properties are going to represent the alpha values in our CSS mask. And since they both default to fully opaque (i.e. 1 ), the entire image shows through the mask. Here’s how we can rewrite the mask using the custom properties:

/* Omitting the @property blocks above for brevity */

img {
  mask:
    linear-gradient(rgba(0,0,0,var(--c-0)) 0 0) left,  /* 1 */
    linear-gradient(rgba(0,0,0,var(--c-1)) 0 0) right; /* 2 */
  mask-size: 50% 100%;
  mask-repeat: no-repeat;
  transition: --c-0 0.5s, --c-1 0.3s 0.4s;
}

img:hover {
  --c-0:0;
  --c-1:0;
}

All we’re doing here is applying a different transition duration and delay for each custom variable. Go ahead and hover the image. The first gradient of the mask will fade out to an alpha value of 0 to make the image totally see through, followed but the second gradient.

More masking!

So far, we’ve only been working with two linear gradients on our mask and two custom properties. To create a tiling or fragmentation effect, we’ll need lots more tiles, and that means lots more gradients and a lot of custom properties!

SCSS makes this a fairly trivial task, so that’s what we’re turning to for writing styles from here on out. As we saw in the first example, we have a kind of matrix of tiles. We can think of those as rows and columns, so let’s define two SCSS variables, $x and $y to represent them.

Custom properties

We’re going to need @property definitions for each one. No one wants to write all those out by hand, though, so let’s allow SCSS do the heavy lifting for us by running our properties through a loop:

@for $i from 0 through ($x - 1) {
  @for $j from 0 through ($y - 1) {
    @property --c-#{$i}-#{$j} {
      syntax: "<number>";
      initial-value: 1;
      inherits: false;
    }
  }
}

Then we make all of them go to 0 on hover:

img:hover {
  @for $i from 0 through ($x - 1) {
    @for $j from 0 through ($y - 1) {
      --c-#{$i}-#{$j}: 0;
    }
  }
}

Gradients

We’re going to write a @mixin that generates them for us:

@mixin image() {
  $all_t: (); // Transition
  $all_m: (); // Mask
  @for $i from 0 through ($x - 1) {
    @for $j from 0 through ($y - 1) {
      $all_t: append($all_t, --c-#{$i}-#{$j} transition($i,$j), comma);
      $all_m: append($all_m, linear-gradient(rgba(0,0,0,var(--c-#{$i}-#{$j})) 0 0) calc(#{$i}*100%/(#{$x} - 1)) calc(#{$j}*100%/(#{$y} - 1)), comma);
    }
  }
  transition: $all_t;
  mask: $all_m;
}

All our mask layers equally-sized, so we only need one property for this, relying on the $x and $y variables and calc():

mask-size: calc(100%/#{$x}) calc(100%/#{$y})

You may have noticed this line as well:

$all_t: append($all_t, --c-#{$i}-#{$j} transition($i,$j), comma);

Within the same mixing, we’re also generating the transition property that contains all the previously defined custom properties.

Finally, we generate a different duration/delay for each property, thanks to the random() function in SCSS.

@function transition($i,$j) {
  @return $s*random()+s $s*random()+s;
}

Now all we have to do is to adjust the $x and $y variables to control the granularity of our fragmentation.

Playing with the animations

We can also change the random configuration to consider different kind of animations.

In the code above, I defined the transition() function like below:

// Uncomment one to use it
@function transition($i,$j) {
  // @return (($s*($i+$j))/($x+$y))+s (($s*($i+$j))/($x+$y))+s; /* diagonal */
  // @return (($s*$i)/$x)+s (($s*$j)/$y)+s; /* left to right */
  // @return (($s*$j)/$y)+s (($s*$i)/$x)+s; /* top to bottom */
  // @return  ($s*random())+s (($s*$j)/$y)+s; /* top to bottom random */
  @return  ($s*random())+s (($s*$i)/$y)+s; /* left to right random */
  // @return  ($s*random())+s (($s*($i+$j))/($x+$y))+s; /* diagonal random */
  // @return ($s*random())+s ($s*random())+s; /* full random*/
}

By adjusting the formula, we can get different kinds of animation. Simply uncomment the one you want to use. This list is non-exhaustive — we can have any combination by considering more forumlas. (I’ll let you imagine what’s possible if we add advanced math functions, like sin(), sqrt(), etc.)

Playing with the gradients

We can still play around with our code by adjusting the gradient so that, instead of animating the alpha value, we animate the color stops. Our gradient will look like this:

linear-gradient(white var(--c-#{$i}-#{$j}),transparent 0)

Then we animate the variable from 100% to 0%. And, hey, we don’t have to stick with linear gradients. Why not radial?

Like the transition, we can define any kind of gradient we want — the combinations are infinite!

Playing with the overlap

Let’s introduce another variable to control the overlap between our gradient masks. This variable will set the mask-size like this:

calc(#{$o}*100%/#{$x}) calc(#{$o}*100%/#{$y})

There is no overlap if it’s equal to 1. If it’s bigger, then we do get an overlap. This allows us to make even more kinds of animations:

That’s it!

All we have to do is to find the perfect combination between variables and formulas to create astonishing and crazy image fragmentation effects.