That button does two things: Continuously Rasterize and Collapse Transformations.
To understand what they mean when they say it continuously rasterizes, it might help if you understood the difference between bitmap and vector graphics.
A bitmap graphic is a picture such as a JPGs and PNGs or a video file such as AVIs and MOVs. It's made up of little colored pixels. Zoom in really close and you can see that.
A vector graphic is not like that. If you go into Illustrator and zoom in really close on something you created there, you'll notice that you never get tiny little squares of color. That's because vector graphics (such as text, shape layers, masks, Illustrator layers, etc.) aren't made up of dots; they are made up of mathematical equations. "This curve goes this far at this angle and is filled with this color," is what the computer knows for vector graphics versus "This dot is this color".
Now, when you import an Illustrator document into AE and place it into your comp it rasterizes because AE is creating video. Rasterization is the process of turning a vector graphic into a bitmap. AE outputs bitmaps. (Usually. I mean, you can do Flash, but that's another discussion for another time.) So, since AE rasterizes your Illustrator artwork, when you scale it up you'll see those nasty scaling artifacts of blockiness/blurriness. However, if you press the Collapse Transformations/Continuously Rasterize switch, instead of rasterizing once, AE does it continuously. Meaning that you can scale your Illustrator layer up as much as you want and it'll remain sharp and beautiful.
Similarly, this also comes into play if you're using a precomp that has had a bitmap layer scaled down inside it and you scale the precomp up. Although, technically, this is due to Collapse Transformations rather than anything to do with rasterizing. The help page explains it thus, "Collapsing transformations can, for example, preserve resolution when a layer is scaled down by half in a nested composition, and the nested composition is scaled up by a factor of two in the containing composition. In this case, rather than performing both transformations and losing image data in the process, one transformation can be performed—doing nothing, because the individual transformations cancel each other.".
Another function is having 3d items in a precomp maintain their 3d-ness in the main comp. This post discusses that.
Honestly, the help file explains it really well, but you do have to understand AE a bit before it'll make sense.
These videos might help you. Part 1 Part 2
(I can't watch them to check because this security wall behind which I work makes some flash videos not play.)
- The Great Szalam
(The 'Great' stands for 'Not So Great, in fact, Extremely Humble')
No trees were harmed in the creation of this message, but several thousand electrons were mildly inconvenienced.