This is an excellent question.
Actually, in my mind, the problem is larger than what you've stated here.
For instance, how should the layers of shadows be combined?
Imagine a scenario where your player is exploring a castle (Super Mario 3 style), and comes accross a portion of the castle which has a hole in the wall. The lights inside the castle create shadows against a ledge, and part of that shadow falls on top of
the hole in the wall. Obviously, the shadow should not be drawn on that portion of the castle, as there is light on the outside of the castle which illuminates the hole. Now, image having multiple paralax layers of the castle, each with different lights and
That may be a bit more complicated that what you're going for, but the problem is still present.
Here's a solution:
1) pre-render the shadows for each layer of the scene onto a single lightmap, sectioned off into different viewports (assuming 4 layers, you could use 4 quadrants of the lightmap texture - one for each layer).
2) Draw each layer, first clearing the alpha channel, drawing the layer, and then multiplying that layer's portion of the lightmap across the whole scene, but doing so only to the amount that is indicated in the alpha channel for
each pixel (ie, draw the shadows only where that layer had been drawn also).
By no means does Krypton directly support this. The Future's portion of the project (available in the latest source) does give more control that will allow for this kind of thing, but there definitely a lot of working going into it.
I may have misunderstood the question. But, I took the liberty of taking notes about how I would implement this kind of functionality, so that in the future I will have them.