Sarah, Part Deux

I decided to restart Sarah Deaner's portrait. Usually I'm against chucking and restarting, so I repaint often and try to persevere when things go badly. There seems to be a bias against doing this-- somehow reworking too much takes away the painting's "freshness"?  I don't know if I agree with that, and certainly x-ray studies of paintings reveal that artists reworked things quite a bit.  The problem with reworking, however, is that with every subsequent layer of paint, the values get darker, so if you want to end up with something that is light in value, reworking might not be the best strategy. In the painting of Sarah, I wanted to blow out the value spectrum and exploit the range as much as possible, but the range was shrinking with every subsequent rework.  Why did I need to rework? I wasn't very happy with the drawing, and I wasn't working from the model at the painting stage, but rather working on colors from my imagination, and I rushed in without a good conception of what I wanted the colors to look like (a color study would have been useful before diving in), which meant I wasted a lot of time pfaffing and rethinking. Anyway, here is the new underpainting, which I started also in a different way which I hope will be more interesting than what I was trying to do before.