I've been attempting to implement tiled WFC, but I'm having trouble differentiating between the different flavors of the algorithm. When it comes to using tiles instead of pixels, most articles (including the github) use the Simple Tiled Model, in which N (size of patterns) is only 1x2. When using WFC in terms of pixels, N is usually 2x2, 3x3, or larger, which is what makes the algorithm so special. So, my question is: Will using an N greater than 1x2 yield desirable results when using tiled WFC?

  • \$\begingroup\$ What structure do you want to capture that is not currently being well-represented with a 1x2 match? \$\endgroup\$ – DMGregory Mar 2 '20 at 21:38
  • \$\begingroup\$ @DMGregory 1x2 could work, but I am curious about whether or not using a larger match is possible, and if so would it yield greater variety and resemblance to the sample material. \$\endgroup\$ – Miles Turin Mar 2 '20 at 21:52
  • 2
    \$\begingroup\$ Try showing us a sample of the output you're getting, and highlighting what parts/traits you're unsatisfied with. Then users can suggest ways to improve those aspects, whether the solution is changing the pattern size, or some other method. In general, you'll get better answers to "I have this problem, how to I solve it?" than "I have this proposed solution, is it possible/beneficial for some problem?" \$\endgroup\$ – DMGregory Mar 2 '20 at 22:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.