For the set up:
I usually do 2, low-res layers for the sketches (because I am using a convertible tablet pc, the sketching is a little
more imprecise than on a wacom), one layer for the lineart foreground (basically tracing the sketch and then often
deleting the sketch layers, this is usually 600 dpi), one layer for the lineart background and then two low-res layers
two work on the shadows for the picture. I tend to not use any special layers for there shadows. I save as
xxx1.oci.
I then work on the basic foreground colors, all of which will wind up having 3 layers each (base, highlight and shadow
which in opencanvas are (as I recall): normal, lighten and multiply; the usual); these are the colors with which I am
most comfortable (often the flesh tone, eye, hair and clothing color). Anything more complicated (complex clothing,
metalic colors) I will color in roughly as well. At this stage I want to wind up with a half-completed picture where the
easy stuff is 90% finished and the hard stuff is just skectched in so I can get a general sense of the final and make
any changes _now_ instead of further down the road, where changes could get really hairy. I then do the same for the
background, but generally, unless there is a _lot_ going on I reserve only three layers for the whole thing (on
occasion, where there are certain particularly tricky areas I will create special shadow and highlight layers for just
that area, which I will merge into the normal background layer before working on anything else). I usually wind up with
two more saves here (xxx2.oci has just the foreground as finished as it will get in this stage and xxx3.oci has the
background finished as well).
The last stage varies greatly from picture to picture, so I can't really comment on that. I do what I have to get
the picture finished.