What does the scaleZ() CSS transform function do?

Paul D. Waite picture Paul D. Waite · Oct 19, 2011 · Viewed 8.4k times · Source

I’m trying to wrap my tiny brain around 3D CSS transforms, and I‘m having trouble understanding what the scaleZ() function is for.

scale(), scaleX() and scaleY() make sense: they stretch the element along the axis specified, multiplying its dimension along that axis by the number you provide.

But scaleZ() seems different:

  1. It applies to children of the element
  2. It doesn’t affect the dimensions of the child elements, as HTML elements don’t have any dimension along the z-axis (i.e. you can’t make a <div> “thicker”).

The WebKit blog says:

[scaleZ()] affects the scaling along the z axis in transformed children.

I can’t figure out what this actually means in practice. Could anyone supply an explanation, preferably with some example code?

Answer

webinista picture webinista · Feb 21, 2013

Seems silly to answer a question two years after it was asked, but posting it for posterity.

It has to do with transform matrices and matrix-vector multiplication. Basically, the transform won't appear to work unless the math works out to produce a product where the Z coordinate is greater than zero.

This is easy to explain, but a little bit hard to understand if you don't have the math background. (But a weekend's worth of WikiPedia reading and Google searches will teach you enough. That's how I learned it.) Every transform function, except matrix() and matrix3d() have an equivalent matrix value. For scale3d, the matrix is:

[sx 0 0 0]
[0 sy 0 0]
[0 0 sz 0]
[0 0  0 1] 

Where sx, sy, and sz are the factor for scaling about the x, y, and z axes. For scaleZ, it's the same, but sx and sy are both 1.

When you apply a transform, the browser takes the coordinates for each vertex of the object (in non-nerd speak: takes the coordinates for each box corner) and multiplies it by the transform matrix. The product of this becomes the new coordinates for the object. For example the math of transform: scaleZ(3) on an object starting at (100,50,0) looks a little like this:

[1 0 0 0]   [100]   [100]
[0 1 0 0] * [ 50] = [ 50]
[0 0 3 0]   [  0]   [  0]
[0 0 0 1]   [  1]   [  1]

That product, [100 50 0 1] when translated into a 3D coordinate system becomes what we started with: (100,50,0). The result is that it looks like the transform wasn't applied. In order for a transform using scaleZ() to have an effect, the third number in the product of the matrix and vector must be greater than zero. It usually happens when scaleZ() or scale3d() are applied to the parent element, or used in combination with another transform function. In both cases cumulative/current transform matrix results in a Z coordinate whose value is greater than zero. Here's an example using transform: rotateY(30deg) scaleZ(3). First we need to multiply the matrix for rotateY(30deg) by the matrix for scaleZ(3).

[0.866 0 -0.499 0]   [1 0 0 0]   [0.866 0 -1.497 0]
[0     1  0     0] * [0 1 0 0] = [0     1  0     0]
[0.499 0  0.866 0]   [0 0 3 0]   [0.499 0  2.598 0]
[0     0  0     1]   [0 0 0 1]   [0     0  0     0]

Then we can multiply our matrix product (that bit to the right of the equal sign) by our point at (100,50,0).

[0.866  0 -1.497  0]   [100]   [86.6]
[0      1  0      0] * [ 50] = [50  ]
[0.499  0  2.598  0]   [  0]   [49.9]
[0      0  0      1]   [  1]   [ 1  ]

Our product [86.6 50 49.9 1] works out to coordinates of (87, 50, 50) if we round off to whole numbers. And that second 50 is our Z coordinate. The transform has a noticeable effect.