But units do rotate along axes other than z, due to uneven terrain, or banking to turn.
Not saying hit cylinders are bad, just that they are not as superior in this engine as you might think.
hit cylinder
Moderator: Moderators
I'd say give us static cylinders now, bother with rotation later. The cases where static cylinders would be less useful than spheres are few (and in those you could still use a cylinder where height~=radius) while the benefits are obvious: Being able to use height as an active part of the units strategic value. Think about it, in real life tanks are built to be as flat as possible so they expose less when taking cover behind a wall or terrain feature. This would give vehicles an edge over bipedal mechs like kbots. In mods where that isn't intended the mechs might need increased hitpoints, although that wouldn't be too realistic either (more moving parts = more space taken up by vulnerable joint assemblies = weaker average armor strength) it should make up for the difference. Also you could have vehs in front of the bots with the bots able to shoot over the vehs and the vehs soaking up the brunt of the damage.
Besides, a hit cylinder wouldn't require any change to s3o as you can just use its radius and height values.
Besides, a hit cylinder wouldn't require any change to s3o as you can just use its radius and height values.
, it sucks for units, since they are tilted. But buildings aren't. But if buildings already use boxes, this idea is obsolete.
I'm pretty sure just the groundplate or yardmap are used for mobility collisions, and the spheres are only used for projectile collisions. In such a case, collision-bodies could be manipulated any which way.FireCrack wrote:That's excelent...
But what is used for unit-unit colossion then?
Imho, the best solution is FBI-defined collision spheres, where the modder can define as many spheres as he like and the tags be (1) anchor part, (2) radius, and (3) offset from anchor.
Thus, by stacking 2 or 3 spheres on top of each other (or orienting them in a square) you could model other shapes vaguely. Woudlnt' be perfect, but it woudl be better than trying to fit every unit into a single sphere.
It would be only used to buildings, so rotation doesnt matter.
Where were these used anyways? Ive spotted that the circles tells where they can shoot, like my laser cant shoot through some other circle because they tries to avoid killing each other.
So are these circles used only for checking bullet collisions faster? (spheres, circles, whatever.)
Where were these used anyways? Ive spotted that the circles tells where they can shoot, like my laser cant shoot through some other circle because they tries to avoid killing each other.
So are these circles used only for checking bullet collisions faster? (spheres, circles, whatever.)
How often would the rotation of a vehicle change enough to make the hit cylinder grossly inaccurate? Few vehicles can climb steep hills so they would rarely exceed the cylinder. I'm not against implementing rotation at all but I think that has time until a later date because it makes the whole affair vastly more complex. Remember that the cylinder is still only an approximation and almost nobody would notice that it doesn't rotate (and if your unit rotates often just make the cylinder as high as it is wide).Dragon45 wrote:Rotation not common enough to matter? You're confusing rotation along certain axes with rotation in general. With a unit climbs a steep hill, its hitcylinder will need to rotate about the Y axis to compensate for the climb
what about during dogfights? planes spin all the time.
Calculating the rotation isn't that easy and requires a lot of math (and will possibly take a performance hit). With a non-rotating cylinder you always know how to treat the axises and it's pretty fast, too. I don't think any other game bothers to rotate cylinders.
Considering how much Spring already slows down with simple sphere collisions I wouldn't want to see what happens if it has to calculate a rotated cylinder for each (an axis aligned cylinder is about as difficult to calculate as a sphere). As I said, just make height=width on your units if you're worried about cylinders, that'll give them a roughly sphereish collision shape.
Considering how much Spring already slows down with simple sphere collisions I wouldn't want to see what happens if it has to calculate a rotated cylinder for each (an axis aligned cylinder is about as difficult to calculate as a sphere). As I said, just make height=width on your units if you're worried about cylinders, that'll give them a roughly sphereish collision shape.