tag:blogger.com,1999:blog-1272803659321539598.post8433774711381269176..comments2024-03-28T23:38:41.403-07:00Comments on Digesting Duck: Navmesh Height Accuracy pt. 2Mikko Mononenhttp://www.blogger.com/profile/11900996590678707801noreply@blogger.comBlogger7125tag:blogger.com,1999:blog-1272803659321539598.post-27130851502446487872009-08-11T23:30:20.029-07:002009-08-11T23:30:20.029-07:00looks like best solution.. callbacks needs additio...looks like best solution.. callbacks needs additional trace to collision geometry, maybe it will slow with complex dynamic environmentYakovhttps://www.blogger.com/profile/12239884265114495349noreply@blogger.comtag:blogger.com,1999:blog-1272803659321539598.post-4772024140729683212009-08-11T23:12:04.886-07:002009-08-11T23:12:04.886-07:00Yakov, the tesselation will be available as separa...Yakov, the tesselation will be available as separate detail mesh for height queries or you can also tesselate the navigation mesh. Usually (especially for games) it is better to have the detail mesh for height only in order to have better performance as you said, but in certain cases you may want to associate some extra data per polygon and more finely tesselated navmesh might be more useful.<br /><br />I will add support for both as it is not really a big deal to implement them.Mikko Mononenhttps://www.blogger.com/profile/11900996590678707801noreply@blogger.comtag:blogger.com,1999:blog-1272803659321539598.post-32813441036607488832009-08-11T22:08:34.721-07:002009-08-11T22:08:34.721-07:00I personally prefer callback approach. But reacast...I personally prefer callback approach. But reacast merge navmesh polygons ... if it will be possible to dont merge voxels with different callback, it will be ideal case imho. Tesselation will slow down A*. For example, on heightmap without obstacles we need only one polygon per tile, but with tesselation it will be very dense mesh. Or you want to have to data sets, one for A*, second for query height?Yakovhttps://www.blogger.com/profile/12239884265114495349noreply@blogger.comtag:blogger.com,1999:blog-1272803659321539598.post-6764197077622562272009-08-11T02:52:04.112-07:002009-08-11T02:52:04.112-07:00http://forums.aigamedev.com/showpost.php?p=20334&a...http://forums.aigamedev.com/showpost.php?p=20334&postcount=38 - about my dynamic navmeshes with RecastYakovhttps://www.blogger.com/profile/12239884265114495349noreply@blogger.comtag:blogger.com,1999:blog-1272803659321539598.post-64063309373641839132009-08-10T12:48:52.290-07:002009-08-10T12:48:52.290-07:00Are you going to check in the current version as t...Are you going to check in the current version as trunk? I have some ideas about the mesh simplification.YAYhttps://www.blogger.com/profile/04242291781994146761noreply@blogger.comtag:blogger.com,1999:blog-1272803659321539598.post-74126614665777271092009-08-10T08:24:27.318-07:002009-08-10T08:24:27.318-07:00The compact subdivision data for the larger level ...The compact subdivision data for the larger level example takes 126kB, where the texture version of the same scene took around 128kB. Some scenes require a little less data. Generating the tesselation is really fast. I could not speed up the point location as much I hoped.<br /><br />I'm not too thrilled on writing a mesh simplification code, but I think I will give your idea a try. The subdivision data is quite tricky, and I would not mind having much simpler data at slight extra cost.Mikko Mononenhttps://www.blogger.com/profile/11900996590678707801noreply@blogger.comtag:blogger.com,1999:blog-1272803659321539598.post-55235253567266652802009-08-09T18:30:23.742-07:002009-08-09T18:30:23.742-07:00Hey great work!
"32 bits per iteration" ...Hey great work!<br />"32 bits per iteration" means 3 chars and a float scaling factor at each navigation-node?<br /><br />I still don't think the extra data is that much after the a lot of sub polygons are merged together (with an given error tolerance).<br />Have you tested it?YAYhttps://www.blogger.com/profile/04242291781994146761noreply@blogger.com