With multi processes you can modify the layer heights, and that's cool. With some careful consideration you could take vertical walls and print them thicker, then lower layer heights for higher angles. Think of a silo. Where it gets round at the top it would make a better print to make the layers thinner as it get closer to the top. However, what about more complex shapes? Sure, you could just slice it thin and have the print take forever, or you can analyze it by hand and make a hundred processes.
I'd love to have an option for the layer heights to be dynamically calculated. Now I recognize the challenge this will be. First of all the calculation will have to take a look at all the normals that cross the slicing plane, and figure out the one least parallel to the build surface. But not just that cross the slicing plane,
we'll also have to take a look at all the faces that we may encounter in the process of drawing that layer. (And I'm guessing that in the odd case that a face that is parallel to the slicing plane will cause some sort of problem. Either that or a perfectly vertical surface with a parallel normal.) Also, normals off parallel with the build plane in the positive or negative direction needs to be considered the same, in order words parallel with the build plane is zero, and everything is taken absolute value from that.
Next, the amount of adjustment of layer height will have to be calculated against the max and min potential layer heights. But is the best result achieved by a linear scale or a curved scale. (My guess is a cosine curve will have something to do with it, just based on my experience with lighting shaders, but I may be wrong.)
Then, how descreet do we want to make the steps? Let's say theoretically, we set the minimum layer height to 0.06mm and the maximum layer height to 0.27mm. If we have a layer that's just a little off vertical, so we want a layer height close to 0.27mm, do we want to do a 0.2699995 layer, or do we want to define specific steps (for those who have done the math and figured out how their Z motors steps and their lead screw calculate into layer heights)?
It's a non-trivial problem, but could result in a considerable savings in time of prints without a loss of detail.