-
-
Notifications
You must be signed in to change notification settings - Fork 1.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Reuse parts of outcome as constraints, and still obtain the same outcome? #66
Comments
Please help me understand your question. By seed you mean a PRNG seed or seed as the constrained part? |
The The problem is weight probability has changed due to constraint propagation. So we can no longer obtain the same tile observation result. My StackOverflow question might be better phrased: https://stackoverflow.com/questions/62490854/order-independent-weighted-random-selection |
I see. So you're trying to get the same result from 2 different constrained parts. Just interesting, why would you want to do this? You can fix the order of observation. Or you can generate result from the first constraint, save this result, and use the choices from the save when generating from the second constraint. |
Imagine this workflow: generate a texture, pick a part of it as constraint, use them to generate again. Preferably, you want the 2nd result to remain the same, and then you change the seed to get different output. But in reality the 2nd result won't match, because constraints reduce entropy, so we cannot obtain the same result even though we know it's one of possible solution. So yeah, it looks like I need to save some more information. |
I see. Yeah, the simplest solution would be to just save the first result, or save the order of observation. |
WFC allows us to set some initial constraints to limit outcomes.
One thing I would like to do, is to use parts of an outcome as constraint, then generate again.
The only problem is, given the same seed and parts of its outcome as constraints, I cannot obtain the same outcome again, because observation order is now different (initial entropy are different).
Can I satisfy this requirement without modifying WFC's underlying algorithm? I think I just need a smart way to do weighted randomization at observation step?
(This is assuming the random input is order-independent, for example, if we use a seeded hash function instead of random number generator.)
The text was updated successfully, but these errors were encountered: