In graphic design circles, a "knockout" is a photographic or artistic image in which an object is digitally separated from its background so that it can be placed against white, black or any solid color, or against any other background image, such as a gradient, a texture, a pattern, or even a separate photo or artistic image.
When such an image is placed in a desktop publishing program such as Adobe InDesign, one has the option of stipulating that text placed adjacent to the image will wrap around the irregular contours of the image, rather than wrapping around a square or rectangle which encloses the image. This is known as a "runaround". So if you want to create a "runaround" (which you most certainly will want to do if you love Dion DiMucci's old song "Runaround Sue"), you'll either need a vector image (such as one might create in CorelDRAW! or Adobe Illustrator or Freehand), or you'll need to create a photo with its own "clipping path" (also known as a "knockout") in Photoshop or another program capable of creating such clipping paths.
All runarounds require the creation of knockouts (except for those involving vector drawings), but not all knockouts are suitable for the creation of clipping paths. A clipping path is essentially a vector shape which delineates the boundaries of the object. But vector shapes, by definition, can't have soft edges (although one can simulate the look of such edges by creating gradients from within vector objects). A knockout, on the other hand, is any separation of the object from its background.
In Photoshop, there are multiple ways to create "knockouts". One way is to create a selection around an object, using one of the many selection tools, including painting tools which can be used in QuickMask mode or on Layer Masks (either of which which can later be saved as Alpha Channels, so that they can be applied during future editing sessions even after objects have temporarily been deselected, and so that they can also be applied to other Photoshop files). Now, those painting tools include soft-edged tools such as the Airbrush tool.
The thing is, you can convert such selections to vector objects (known as "paths" in Photoshop), and you can use those paths in order to create exportable clipping paths which can be used in programs such as InDesign for text runarounds, but they won't retain their soft edges.
One can adjust the fidelity of such conversions from selections to paths, in order to increase or reduce the total number of Bezier control points in the path. The more control points there are, the more likely it is that the resulting vector paths will faithfully replicate even the most intricate details in the original selections. That may or may not be a good thing. If you have a somewhat shaky hand when you're drawing, then creating a vector shape with a reduced number of control points can be one way to create smoother lines and curves. You may have to edit the control points slightly afterwards, though, in order to cause them to conform to the original shape as much as possible. The fewer control points there are, the greater the likelihood that the resulting shape won't precisely follow the original shape.
You should also know that vector graphics and raster graphics are different in terms of how their contents affect file sizes.
All digital photos and scanned images are considered to be "raster" graphics. The files contain detailed information about the specific pixels in the image. The file size of raster graphics is determined by the resolution and bit depth of the file, and also by whether or not compression is used, and by the type of compression and the settings used when compressing the file.
With vector graphics, such as paths, the actual number of pixels is continuously variable and infinitely scalable, since the digital file doesn't actually store information about actual pixels. Instead, it stores instructions pertaining to shapes, and the contents of those shapes, and then the actual number of pixels is determined by the settings used for displaying or printing the image and by the specific characteristics of the display device or the printing device.
Generally, vector graphics are far more efficient than raster graphics in terms of data storage, because it takes far less data to describe a shape and its contents than it takes to describe every individual pixel of which that shape is comprised. But whereas the file size of raster graphics is determined by the factors I've previously mentioned, the file size of vector graphics is determined, in part, by the number of control points. So simple vector shapes take up less data space than complex vector shapes. It is possible, using some sophisticated raster-to-vector conversion programs, to create vector illustrations which will almost (but not quite) fool you into thinking that you're looking at a photo or raster illustration. But it should go without saying that such images feature far more control points than simpler vector illustrations. They still retain certain advantages over their raster counterparts, in terms of infinite scalability, but their advantage in terms of file sizes is likely to be somewhat diminished.
It's also possible, in some cases, to produce vector illustrations which are so complex that their code causes output devices or computers to crash when attempting to perform the vector-to-raster conversions which are necessary for printing purposes. It's important to be sure that one's RIP workstation has sufficient memory for such purposes when printing such illustrations. Obviously, that isn't nearly as much of an issue as it used to be, since it isn't uncommon to see computers with several gigabytes of RAM, but it's still something to be aware of.
If you're a musician (as I am), you may compare raster graphics to digital audio files (where the digital code represents and then recreates the actual sounds), and you may compare vector graphics to MIDI files (where the code merely stores instructions which are then interpreted by the output device, such as a digital keyboard, or a MIDI sound module such as a sampler, or a "virtual instrument" which streams from a computer's hard drive during playback).
Now, the most typical use of "knockouts" may be for commercial applications, where products being offered for sale via catalogs or flyers or e-commerce websites are separated from their original backgrounds and then placed against solid white backgrounds or other similarly simple backgrounds which make it much easier to focus on the products themselves without being distracted by superfluous details.
Vector paths are particularly well-suited when it comes to creating such knockouts, because many products such as electronic devices, cars and so forth lack the soft edges which might be seen in portraiture and other images of that type. But there are exceptions. Teddy bears, for instance, are soft and fuzzy. It's unlikely that a vector selection around a soft and fuzzy teddy bear would look as good as a raster-based selection, if one were to closely inspect the edges of the image. But even in such cases, the requirements of commerce aren't usually as demanding as the requirements of fine art, so it's possible to use paths for such purposes and still get an acceptable result.
If you're accustomed to creating vector graphics and working with Bezier curves and control points on a regular basis, then creating paths in Photoshop ought to be easy. But others, such as myself, tend to prefer raster based tools because they feel more like traditional media such as paintbrushes and pencils. If one only works with vector objects occasionally, creating the appropriate paths is possible, but it can be time consuming.
That's why it's good to know about ProKnockout.com, run by a company which does virtually nothing other than to create vector paths for raster objects in Photoshop. Personally, if I were asked to create a large catalog or e-commerce website which required a lot of knockouts, I might be tempted to use their services so that I could focus on other aspects of production.