I hope this is an ok place to put this, but I just had a very interesting idea. Almost all materials have a freeze-thaw hysteresis. I forget what it is called, but even ice has this. Suppose, for instance, it melts at 0.1 degrees, and does not re freeze until -0.1 degrees. So suppose you had a block of transparent, bubble free ice in a very precisely controlled chamber, at very close to 0 degrees. You selectively melt the ice in the relevant places, and since the refractive index of liquid water is almost (exactly?) the same as solid ice, you can point the light source wherever you please. To avoid melting along the path of the light, you can focus it on a point, so it is far higher intensity there than elsewhere.
Once melted, the water remains liquid. After you are done melting out a suitable cavity, drain the water, evaporate the last of the liquid with dry air, and then fill the cavity with epoxy, or plaster, or similar. Let it set, then melt away the remaining ice, and you have a nice object.
Now, this example is with water so we can all imagine how it might go easily enough, but I'm sure there are better materials for the task. Some materials have a hysteresis of tens of degrees, but we also want it to be transparent, and for the liquid phase to have nearly the same refractive index, ideally, so the light can be directed anywhere in the volume accurately and still come to a very small point, and not get all blurred etc from lensing along the way.
Ideas on what materials would be better than water/ice? I wonder if there is an additive to water that increases the hysteresis, thereby leaving us less burdened by the need to be very precise about how much energy is deposited by the light source. Although really I wonder if ice would actually be ok.