I think - just perhaps - I've got the layout for the dormer on a bias sorted now, though with some small variations from the text's example. The text shows the valley pieces, noulet, being joined at their peak with a form of birds mouth joint. I reconfigured it so that the pieces would join with a miter, which means that one of the valley pieces cannot come from a square section stick of wood. The valley pieces are shaped as trapezoids in cross section, so that they can simultaneously lie flat on the main roof and, on their other sides allow a common rafter to join to them with a simple (that is, non-compound) miter cut. And of course their top and bottom edges are in plane with the dormer roof surface, both atop the rafters and in plane with the under-surfaces of the rafters.
This is the completed dormer:
On the front is the tympanum molding.
Same view, with roofing boards and main roof plane removed:
The boards which lay down on the roof surface and travel on a bias, serving to receive the dormer wall plate on each side, are parallelograms in section.
Here's a view normal to the front, which is aligned to the edge of the roof:
A snapshot of the 2D drawing fun which produced the complicated parts of the dormer framing - the noulet in particular:
In some cases I needed to reverse engineer from the 3D to determine exactly what was going on in spots. It did all come together in the end.
I then stuck the biased dormer on a drawing with the hip roofed, otherwise orthogonal version:
Plan view of the completed biased dormer:
Last, a view from the other side:
That's two dormers down finally, and the drawing method is starting to make more sense. Maybe I'm kidding myself though. A long way to go yet in this study. Each drawing in the book of a given set of examples, like dormers, is a little more complicated than the previous example. Next up is the lucarneéventail, a fan-shaped dormer with a descending ridge and rotated rather than backed hip rafters. It is definitely the most attractive dormer example in the book to my eyes. I had a go at it about a year ago and ran into some difficulties with the 2D part, though I was able to complete it in 3D. Recent conversations with Tim Moore in France though have allowed me to see some new things in that drawing and I feel like I should be able to surmount the previous problems. I'm sure in collaboration we can get it done.