Creating a UI – not sure if this is how others do it, but the way I work is to have the overal base set on 1 page. Give it the look and feel that it needs in the end or at least on a very basic level.
“This goes here, its about this size, it needs to have an element that is this size next to something like this with this radius, with this type of a menu to the side. Oh, need the the little arrow thing indicating a drop-down. etc.”
There’s more to it, like the orientation of how its put together – is it going to expand downward when “adding” a new item? Ok, then stack panel, is it sticking here no matter what? How is that button going to look when its hovered over, active, or inactive?
All of these thing are the beginnings of something more… whole. Once they’re together at least in a rudimentary sort of form, custom controlling the… (insert word of choice) out of it until there’s almost othing left on the original screen.
From there – its detailed creation time. Inactive, mouse-over, and active alike. Personally, I like to also make these things “work” – as in, opening Visual Studios and getting the custom control to behave when running my mouse over it and clicking.
Makes it so that when I present it to the client, it may not actually… transmit, record, and display data, but at least the UI is somewhat interactive.
Problems come in though – Hitting F5 from Blend to run the program and if the tiniest thing is off, it either won’t load or… it will but – *growls*
One might say that there is a very very nice use of “white space”.