he second type of image coordination in a design composition, which I will call embedding, is a direct opposite to displaying.  Here, the image has no well-defined boundaries or limits, and the space it occupies only depends on its content: If a figure depicted stretches his hand, the image will stretch as far as the hand does, or just a little bit further.  The way the image blends into the background depends on its dominant texture; for example, for the blurry strokes of the Oracom artwork, blurring the edges worked great, while some more complex and realistic textures would have to use gradually diminishing strokes connatural to those of the image interior.

Sometimes an image is clipped out of the background at the borders of an object, just as one would do with a photo.  Even with a photo that is a bit out of focus (or simply enlarged beyond a certain limit), a crisp edge would contradict with the blur of the interior.  It therefore rarely works with artwork whose wealth of textures makes any "clear cut" look artificial.  In other words, if you want a piece of artwork to be masked off of the background, you should paint it that way, because cutting the "live flesh" of a completed picture is painful and often unsatisfactory.  (On the contrary, the displaying method above works best with crisp linear edges, regardless of the image's texture).

Thus, it is easy to figure out what sort of images could benefit from embedding: namely, those containing a single integral object with more or less obvious borders, where it is easy to tell the object from the surrounding background or there is a pronounced compositional center.  Due to this inherently centric nature, an embedded image usually acquires a significant compositional power on the page, sometimes taking control over the rest of the elements and arranging them relative to itself (as it happened on the Oracom brochure's first page where the image occupies the dominant central position).

If we put such an image, with its own intrinsic opposition of the center and the periphery, into a rectangular frame, that would introduce another "layer of bounding" resulting in a clash or, at least, a tautology.  Therefore reusing, or creatively elaborating upon, the contours already present in such images seems to be the only satisfactory way for implanting them into design compositions.

An example is presented on Fig. 4: An advertisement page for a fictitious dance school uses a dancer picture which, like most other monofigure works, is embedded in the composition.  However, since the texture of this image is quite interesting by itself, its small, open-ended fragment with no recognizable objects on it could be displayed as a separate decorative element in another composition (Fig. 4, bottom).  Conversely, a multifigure displayable composition may provide you with embeddable figures if you'll manage to cut them out of the background separately.  As we've seen in the Oracom example above, the original embedded image can be the source of both embedded and displayed fragments.