No option for tile with no symmetry about wavefunctioncollapse HOT 8 CLOSED

Here is a solution:
(1)Add the following lines to the list of symmetry operations/functions.
else if (sym == 'F')
{
cardinality = 8;
a = i => i < 4 ? (i + 1) % 4 : 4 + (i - 1) % 4;
b = i => i < 4 ? i + 4 : i - 4;
}

(2) Add reflection function after line 47 ("Color[] rotate(Color[] array) => tile((x, y) => array[tilesize - 1 - y + x * tilesize]);"):
Color[] reflect(Color[] array) => tile((x, y) => array[tilesize - 1 - x + y * tilesize]);

(3) Change 135-136 ("tiles.Add(rotate(tiles[T + t - 1])); tilenames.Add($"{tilename} {t}");") as:
if (t < 4) tiles.Add(rotate(tiles[T + t - 1]));
if (t > 3) tiles.Add(reflect(tiles[T + t - 4]));
tilenames.Add($"{tilename} {t}");

from wavefunctioncollapse.

Here are some of the tiles that it is meant to connect to:

from wavefunctioncollapse.

Hi! You're right, there is indeed no option for a tile with no symmetry. I'll patch this up eventually, but I'm not sure when, it requires more than just adding else if (sym == 'J') { cardinality = 8; ...; }. I'm releasing a new version of WFC soon (as part of a larger project) with a different symmetry system that doesn't have this problem, because it induces the symmetry type from a low-pixel "scheme" of a tile.

from wavefunctioncollapse.

Hi mxgmn,

Thanks for your reply - I'm glad I understood the problem correctly! Could you explain why adding a "cardinality = 8" option would not fully solve the problem? For instance, adding the following symmetry option does not solve the problem but I do not understand why:

cardinality = 8;
a = i => 0;
b = i => 0;

from wavefunctioncollapse.

Because it currently assumes that all symmetries of the tile could be reached by rotating the base tile (which is not true for your example). And by setting i => 0 you are actually describing a maximally symmetric case. The right a action would look more like i => (i + 1) % 8.

from wavefunctioncollapse.

I'm late to the party, but oh well, at least for posterity: the already mentioned fast-wfc has a symmetry="P" option, and additional indices for flipped+rotated tiles (4 for horizontally mirrored tile, for example)

from wavefunctioncollapse.

how is this related to chaos theory? & hope for the demo!

from wavefunctioncollapse.

I tested @chriskkkkl's solution on the new tileset, it works.

from wavefunctioncollapse.