tag:blogger.com,1999:blog-557402264081468948.post6012812756511469359..comments2016-10-02T01:42:27.164-07:00Comments on Computational Bubblegum: Implementing Continuum Crowds, Part 2: FieldsDani Phyenoreply@blogger.comBlogger3125tag:blogger.com,1999:blog-557402264081468948.post-91757248634758297302015-02-11T14:08:49.476-08:002015-02-11T14:08:49.476-08:00Thanks :) The best resource for explaining that is...Thanks :) The best resource for explaining that is the "Implementation Details of the Fast Marching Methods" in the Fast Marching Method Wikipedia page. I'm mostly referring you to there because, for anyone that is implementing this seriously, the method I described here can be slightly off, and they describe this method here first, and then a better method that's also more complicated to implement. They also describe why both methods work, which I didn't explain here either.<br /><br />That aside though, if you get two solutions from your quadratic, you want the Max(solution1, solution2), not the min. I fixed my scicomp question accordingly. This is because the current node must be larger than it's neighbors, and the smaller solution won't fulfill this criteria.<br /><br />On your other question, you're very right, sorry. If our density is lower than pMin we should only use the topological speed, while if it's greater than pMax we should only use the velocity field's speed. That is now fixed.<br /><br />To be clear, the method I gave does work for, say, pathfinding. It just isn't "perfect" in computing T(x,y), meaning that if you are depending on the exact behavior of this function for an application you need to tweak this method a little to get as close to the exact behavior as possible.Dani Phyehttps://www.blogger.com/profile/05701735406861236917noreply@blogger.comtag:blogger.com,1999:blog-557402264081468948.post-49855709527377453422014-11-05T04:45:19.396-08:002014-11-05T04:45:19.396-08:00My copy paste didn't succeed...
if ρ(x+1,y+1)ρ...My copy paste didn't succeed...<br />if ρ(x+1,y+1)ρmax then f(x,y)=fT(x,y).guillaumehttps://www.blogger.com/profile/05087475522725842765noreply@blogger.comtag:blogger.com,1999:blog-557402264081468948.post-37581648867590616362014-11-05T04:43:51.045-08:002014-11-05T04:43:51.045-08:00Very interesting topic, and well explained !
I'...Very interesting topic, and well explained !<br />I've noticed some inconsistencies though. In the first part, about calculating T(x,y) it is said that the solution to be considered is Max(solution1, solution2), whereas in the scicomp forum it is the min (I guess it is the min but...).<br />In the second part, I quote:<br />if ρ(x+1,y+1)ρmax then f(x,y)=fT(x,y).<br />Isn't it the opposite ?guillaumehttps://www.blogger.com/profile/05087475522725842765noreply@blogger.com