The crater field's lattice ladder, coarsest first.
Each level halves the diameter and doubles the cells per axis, which quadruples the cell count — so one crater per cell reproduces a cumulative size–frequency slope of −2 for free, which is the lunar highlands at saturation (Robbins 2018). Nothing has to enforce the slope; it is the geometry.
The production slope is steeper than −2, and that is where density
comes in: an unsaturated surface is thinned at the large end and approaches
saturation at the small end, because small craters accumulate faster than
they are destroyed. So the per-level chance climbs toward 1 as the diameter
falls, and a young surface has a steeper distribution than an old one rather
than the same one scaled down.
The ladder is a property of the body, not of the patch, and the obvious
alternative is what makes that worth stating. Stopping the ladder at each
patch's own sample spacing would be free detail control — an orbital patch
would not pay for boulder-scale craters — and it breaks the morph. A CDLOD
child hands over by sliding its vertices onto its parent's grid, so a fully
morphed child is the child's own field sampled at the parent's spacing; that
equals the parent's mesh only if both evaluate the same function. Two patches
with different ladders differ by every crater between their two floors, and at
level 12 on a lunar-sized body that is eight meters of pop at the handover.
One field, one ladder, and the cost ceiling is MAX_CRATER_LEVELS.
- grammarSurfaceGrammar
- Undocumented
- floornumber
- Undocumented
- limitnumber
- Undocumented