In my production phase, I engage in what I term a process of assembly.  The goal here is to generate something that “hangs together” well—something that is “…a kind of unitary whole… [wherein] all the ‘parts’ have…some relationship with each other” (Whipple); a piece wherein, I would hope, the piece is at least the sum of its parts, and perhaps more than that.  Once I’ve collected (and pre-generated) enough material—that is, a quantity sufficient for me to mix and match and have a chance of something coherent coming out (and here I go with a gut feeling, rather than a firm plan), I select and assemble the items I will perhaps use.  Again, text comes first, then media.  This bothers me a bit, for I find myself wondering if I am not letting the media “have its own”, so to speak.  And my response is that no, I probably am not; again, this is a product of my cognitive bind to traditional text, and, probably, my familiarity and relative experience in creating multimedia pieces in the first place, as well, I am sure, of my familiarity with the various possible tools available to record, construct, and manipulate media (e.g., iMovie, MovieMaker 2, iPhoto, Windows Media Encoder, Audacity, et. al). Assembly is a pretty recursive process for me in multimediation, as it is for traditional text. It’s akin, perhaps, to a kind of cookery.  I’ll try something out in one place, add something to it; perhaps I’ll consider an image or piece of media I have created, moving it around, or try out one font or template scheme and replace it with another; but more often I will add text, add explanation, usually justification, like a chef will add spices or salt or more carrots.  My assembly process is not nearly as straightforward as Murray’s “first draft’ seems to imply, at least to me, and take a great deal of time—probably the largest single chunk, or group of chunks, of time in the whole ‘newmediating” process.  For example, during the construction of the text you are reading now I have moved, replaced, deleted, amended, and added, at least half a dozen times, pieces of text that were, in the collection phase, tentatively intended for other sections, either earlier or later than their eventual placement in the text.  
In the assembly portion, I decide where to put media elements based on their relationship to the text. “Relationship to the text” of course, is a rather general indicator.  But I rarely make lengthy decisions when I consider where to put media.  Sometimes the relationship is illustrative; for example, I might put a picture of my smartphone in the text above illustrating what I mean by the term ‘smartphone”, and perhaps showing the lens side of the instrument, to “prove” or illustrate the fact that it can, indeed, take photographs.  As another example, in a multimediated article published in 2004, I included a site map graphic to allow the reader to take her or his own path through a moderately complex piece.  
Likewise, I may make initial determinations as to color, design, font style and size, and the layout of the piece.  I tend to leave layout until the end of the production process; I tend to want to have the content at least partially nailed down before I tinker with the format.  In saying this, by the way, I realize that I am separating design from meaning in a way that I suspect someone either more experienced or more specifically-educated in multimedia design would not do.  It seems that as a multimedia writer, at least, I am not only separating design from meaning, but also separating what could be called “content” from “effect,” with traditional text forming the large part of the former and the media forming the latter.  Again, I ascribe this separation to my techno-rhetorical education, with the “techno-“ following the “rhetorical” by a good dozen years or so.