3 ms·
A Wave Function Collapse Overlapping Model Guide
- ianai 7y agoI broke it. (In a good way) I managed to embed a pattern in the feed that is not repeated in the output. Well, at least some of the time is not repeated-haven’t done very many iterations. https://ibb.co/k6BmFsd https://ibb.co/k6BmFsd
- 42yeah1 7y agoThat might be due to the blue rect and the green rect must be separated with a straight line, and straight lines are hard to come by: https://imgur.com/85W15eI https://imgur.com/85W15eI
- skolskoly 7y agoI broke it in a bad way. Pattern chugged along fine and then it froze and a random white rectangle appeared. (N=3) https://imgur.com/a/IlmdkC6 https://imgur.com/a/IlmdkC6
- dfee 7y agoThis is awkward. My home router blocks your site as malware. (It tells me I can submit the site for review here: bitdefender.com/submit
- 42yeah1 7y agoWell maybe that's because I am hosting my blog on vultr... I guess?
- a-dub 7y agocomputer modern in the hizzy!
- rawoke083600 7y agoWow that's cool ! It's like perlin noise for grownups :D
- symmetricsaurus 7y agoSeems like n=3 really helps to get good results. With n=2 it was a bit hard to see that it actually followed the pattern.
- 42yeah1 7y agoYeah, that's why in the official repo lots of images were generated using N=3. However N=2 is pretty good for mazes and stuffs from my perspective.
- BorisTheBrave 7y agoNote that the implementation described above uses the Arc Consistency 3 algorithm for the propagate step. It's often an order of magnitude faster to use AC4, as is done in the original WFC repo and in my implementation (https://github.com/BorisTheBrave/DeBroglie https://github.com/BorisTheBrave/DeBroglie).
- 42yeah1 7y agoYeah, I am just trying to make it work :P I would strongly advise against using my implementation.