Sure! I actually do some really shitty thumbnails before doing anything. This is one I’m working on right now:
Mostly I figure out what is important first. I know I want some mysterious moonlight falling upon my sinner lady and I want the minister to be creepily behind her, rosary in hand. The first comp went pretty well, but I’m not getting in the dramatic shadows I really want to showcase with that snazzy moonlight, so I move onto thumbnail 2, which is sadly much more difficult to just jot down, so I throw some very loose and probably inaccurate perspective in by starting with the floor as the base of the box and building from there. After this, I sometimes go into Google SketchUp, build some sort of mock up and rotate it around until I get an idea of the perspective being correct. Then I’ll either screenshot that and redraw my sketch overtop the model in photoshop, or I’ll look at it while drawing for reference. And the it’s on to coloring and such. Sometimes I do b+w first, then use gradient maps, overlay, and color layers to color an image, or sometimes I know off the bat what I want my colors to be. With bigger pieces, I’ll go back to the thumbnail and do a color thumb to make sure I’m not throwing in too many colors and my lighting looks unified.
That’s sort of the sparknotes version, but I don’t want to ramble. Let me know if you want to know more!