Enzymind Labs

I think it is straight-forward (but I agree, not obvious), or I mis-understood the question.

Usually people think of these (Hilbert, Peanno, etc.) as composed of curves (sets of infintessimally thin connected line segments), and the L-systems as replacements of each segment between kinks with a new set of segments. In this way the "maze", or space between the curve, is not explicitly represented.

But there is an equivalent way of constructing the curves (in the infinite limit) using a replacement system that represents the intermediate steps as blocks (maze and walls -> symbol 1 and symbol 2 -> white and black squares). All of the L-systems that I am considering are of this form, and it has the advantage that the "maze" portion (between walls -> symbol 1 -> white squares) is explicitly represented. They are also very simple to implement. For example see these two symbol systems, these three symbol systems (only a few of which are space-filling), and these tilings of space-filling curves. The can be manually constructed (and several of mine are) using copy and paste, recursively, in an image editor.

Let me know if I can help you understand how I did these, my documentation is still thin. The more specific the questions the better. Do you have the link where this is mentioned in Cut-the-Knot?