Several can play the selfquote game. box/cylinder for detecting selection and detecting if you need to per polygon collision detect.[K.B.] Napalm Cobra wrote:A combination of box/cylinder and perpolygon would be best.
Map "Features" models, pictures, models, comments,
Moderator: Moderators
-
[K.B.] Napalm Cobra
- Posts: 1222
- Joined: 16 Aug 2004, 06:15
-
Warlord Zsinj
- Imperial Winter Developer
- Posts: 3742
- Joined: 24 Aug 2004, 08:59
You can't make the whole block a single model, because it defeats the purpose of having a city battle in the first place. The things aren't just giant eyecandy, they are there to have an impact on the battle. You should be able to sneak around buildings, and use them as cover, and set traps on the roads, while using the byroutes to get through, etc.
-
smokingwreckage
- Posts: 327
- Joined: 09 Apr 2005, 11:40
To illustrate the point, buildings must eb small and far apart for units to go anywhere near them or inbetween.
Aircraft cant approach buildings.
units can go near but not upto buildings.
Only small untis can go inbetween them.
Pathfinding for these untis must surely be affected.
Selecting units near a building will be a nightmare
-
LathanStanley
- Posts: 1429
- Joined: 20 Jun 2005, 05:16
-
el_muchacho
- Posts: 201
- Joined: 30 Apr 2005, 01:06
I know nothing about collisions, but I wonder how it's possible to calculate collisions with boxes (or even worse, chamferboxes) as easily and quickly as with spheres. With shperes, one simply calculates the distance to the center (i.e sqrt(deltaX^2+deltaY^2+deltaZ^2) ), while with a box, I've yet to figure out the calculation, but it would probably involve calculating a few cos and sin of the angle between each box surface and the incoming missile/laser/wall/whatever. 
Exactly, there are no simple and fast collisions possible on chamferboxes. You might just do full mesh-mesh collisions then.I know nothing about collisions, but I wonder how it's possible to calculate collisions with boxes (or even worse, chamferboxes) as easily and quickly as with spheres. With shperes, one simply calculates the distance to the center (i.e sqrt(deltaX^2+deltaY^2+deltaZ^2) ), while with a box, I've yet to figure out the calculation, but it would probably involve calculating a few cos and sin of the angle between each box surface and the incoming missile/laser/wall/whatever.
Since buildings can't be rotated i'd say axis aligned bounding boxes for buildings (works better with tall buildings), and spheres for units because they're the fastest primitives.
-
LathanStanley
- Posts: 1429
- Joined: 20 Jun 2005, 05:16
point understood... I'm pretty good with calculus.... I'll see what I can do about possibly intriguing some thought on a more "precise" collision detection...el_muchacho wrote:I know nothing about collisions, but I wonder how it's possible to calculate collisions with boxes (or even worse, chamferboxes) as easily and quickly as with spheres. With shperes, one simply calculates the distance to the center (i.e sqrt(deltaX^2+deltaY^2+deltaZ^2) ), while with a box, I've yet to figure out the calculation, but it would probably involve calculating a few cos and sin of the angle between each box surface and the incoming missile/laser/wall/whatever.
-
LathanStanley
- Posts: 1429
- Joined: 20 Jun 2005, 05:16
well if its currently running the roots sqrt(deltaX^2+deltaY^2+deltaZ^2) type collision dection on a sphere....
couldn't you add in the lines: sqrt[delta(R+dx)^2+delta(R+dy)^2+delta(R+dz)^2]
don't get me wrong with dx, dy, and dz... they are not derivatives... they are "numbers" used to represent the hight, width, and length differences..
such a number in the forumla would adjust the sherical bounding box to an elliptical one, covering the endpoints, and generating a "close but not perfect" type box...
and yet would still have a simple formula... with little change needed to be done to the engine..
the actual collision detection would only need a formula change... but the added numbers would have to be defined...
as I understand now, all units are given a SINGLE number determining its bounding collision sphere thingie...
what would have to be done, would be every unit would be given an X,Y,Z bounding numbers origionating from its center location..
example
couldn't you add in the lines: sqrt[delta(R+dx)^2+delta(R+dy)^2+delta(R+dz)^2]
don't get me wrong with dx, dy, and dz... they are not derivatives... they are "numbers" used to represent the hight, width, and length differences..
such a number in the forumla would adjust the sherical bounding box to an elliptical one, covering the endpoints, and generating a "close but not perfect" type box...
and yet would still have a simple formula... with little change needed to be done to the engine..
the actual collision detection would only need a formula change... but the added numbers would have to be defined...
as I understand now, all units are given a SINGLE number determining its bounding collision sphere thingie...
what would have to be done, would be every unit would be given an X,Y,Z bounding numbers origionating from its center location..
example
-
[K.B.] Napalm Cobra
- Posts: 1222
- Joined: 16 Aug 2004, 06:15
-
Warlord Zsinj
- Imperial Winter Developer
- Posts: 3742
- Joined: 24 Aug 2004, 08:59
