Looks like the AI pathfinding will finally be "fixed" in the nearest future: pull request 1633.
If both this and AnyOldName3's shadows get in the release 0.44.0, it's gonna be thrilling.
Two interesting PRs to watch
-
- Posts: 24
- Joined: 30 Aug 2014, 13:19
Re: Two interesting PRs to watch
The pathfinding features appear to be quite significant, it's content like this that will finally lead to fixing bug #1000.
Re: Two interesting PRs to watch
Shadows have been quiet for a while. Hope everything is ok.
- AnyOldName3
- Posts: 2683
- Joined: 26 Nov 2015, 03:25
Re: Two interesting PRs to watch
The thing I'm working on right now requires replacing a whole chunk of code that I kind-of understand, but suspect is thoroughly flawed, with something that works, but the intermediate state is something that works even less well, so there's not much point pushing it to GitHub until it's done. My girlfriend also broke up with me, which took the wind out of my sails for a couple of weeks, so not much got done then.
The two big things now are that sometimes the bounds of light space are becoming far bigger than the view frustum (making shadows look incredibly crap in certain circumstances) and that the whole of the adjustPerspectiveShadowMapCameraSettings function is complicated and potentially broken and wrong and I want to replace it in its entirety. It doesn't help, though, that the demo code for the original LiSPSM paper isn't especially well organized and uses a left-handed coordinate system which sometimes makes no difference to the maths, and sometimes makes cross products wrong.
The two big things now are that sometimes the bounds of light space are becoming far bigger than the view frustum (making shadows look incredibly crap in certain circumstances) and that the whole of the adjustPerspectiveShadowMapCameraSettings function is complicated and potentially broken and wrong and I want to replace it in its entirety. It doesn't help, though, that the demo code for the original LiSPSM paper isn't especially well organized and uses a left-handed coordinate system which sometimes makes no difference to the maths, and sometimes makes cross products wrong.
Re: Two interesting PRs to watch
You are a champion! Good luck!
Re: Two interesting PRs to watch
I'm sorry to hear that I hope you're doing better, now.