Universal construction of complex patterns from 1 cell

[PkmnQ]

How about a rule that prints ITSELF?

You mean a sort of ruleloader quine?

[andrewbayly]

I have been looking at your improved version - BLAHMONALISA2.rule and I'm curious. How did you achieve such a minimal set of transitions. There appear to be roughly 30,000 transitions in your ROM, but there are still 60,000 painted cells in the Mona Lisa.

[blah]

I have been looking at your improved version - BLAHMONALISA2.rule and I'm curious. How did you achieve such a minimal set of transitions. There appear to be roughly 30,000 transitions in your ROM, but there are still 60,000 painted cells in the Mona Lisa.

Sorry for taking a while to respond, for some reason I wasn't subscribed to this thread.

As I remember it, a transition only needs to be specified if the colour changes. When there's no transition, the colour stays the same between subsequent pixels. Many regions of the picture have subsequent pixels of the same colour, ergo there is no need for a transition there. (The same idea applies to the extender signals that push the arm out by one cell to the right; I don't remember if they were already optimised in this fashion in the original version).

Having fewer transitions evidently makes the rule tree simpler and therefore faster to access, I suppose.

P.S.: It's crazy to me that this was already 8 years ago. To my knowledge, my 1-cell-universal-computer-spaceship idea using this mechanism still hasn't been built. Here's to another 2**3 years of not doing that idea.