Sunday, October 2, 2016

Splitting headaches (with supercontinents)

I went up to Dayton today for the Dawn Patrol Rendezvous, and watched a variety of manned replicas of WWI airplanes fly through the skies above the Air Force Museum.  When crosswinds interfered, large-scale RC planes took their place in the skies, and performed more daring maneuvers. I'll have a more thorough report in a future post, but that is why I made only a little progress today on the supercontinent splitting.
Supercontinent 1
Split result 1
To perform an initial split of the supercontinent, I tried a new approach.  I used a fractally-perturbed "vertical" line in the center as a very rough approximation of a mid-oceanic ridge acting as a spreading center.  The result was okay - it seemed like a bit better result than where I was at back on Thursday.  But it wasn't the end result I was looking for.  I ran it again and was even more disappointed.
Supercontinent 2
Split 2
The second run was even more disappointing.It is just too obviously close to vertical.  I'm going to tweak some constants slightly but I expect that to be of only marginal value.  My thought if that doesn't pan out is to switch from a vertical baseline to a somewhat diagonal one, again perturbing it with fractals.

Side note: My initial approach wasn't using Perlin noise, but to a poly line and midpoint displacement.  Performance was very poor, however, and results less impressive still.  Switching to Perlin noise like I'm using on the Planets generator was an improvement, even if I'm still not fully satisfied.

Another approach also occurs: divide the mid-ocean ridge into small number of line segments rather than one line.  This could allow for greater variety without some of the problems I was running into with poly line subdivision and midpoint displacement, while retaining the performance boost of the current approach.

My alternate approach may make you wonder why I don't just use the probablistic fill that generates the plate division.  They already generate more varied splitting lines.

Also, rotating plates slightly could be of benefit, visually.  While the coast lines of South America and Africa fit together almost like puzzle pieces, there is a bit of rotation involved.  Other continents rotate quite a bit more.  Alas, they also have a tendency to produce lots of long, thin pieces.  The plates arrangements shown in Split 1 and Split 2 are not particularly bad in that regards, but do help illustrate the point.

We'll see what tomorrow brings.  Hopefully I'll make more interesting progress on this, and have more on the Dawn Patrol (some photos and a better report would be nice).





No comments:

Post a Comment