Im not dead...

[Min3mat]

heh, i thought u were the admin now, aren't u supposed to know stuff like that?

[AF]

Weapon collision is indeed sphere based, as is mouse selection, unless you drag a box over the unit in question.

[zwzsg]

OH NO IT DOES F***ING NOT!!!

You think me, aGorm, would be lying to you? They do not use the footprint size, NO MATTER WHAT YOU SAY, I should know, becasue I've tested it out!

Anyway, I have teh final say becasue I can quote Zaphod himself:

Jelmer says:
zwzsg is wrong

PS, Jelmer is Zaphods MSN name....

aGorm

BIGGER REDDER FONTS MEANS MY POST IS TRUER!!!!

I can invoke quote from coders under their alt nick too:
[22:59] <zwzsg> Since I'm currently trying to get a model with a large looking model ingame but a minuscule collosion sphere.
[22:59] <zwzsg> So it doesn't collide with the factory it's holding on its back.
[23:00] <zwzsg> Can you tell me how it's calculated?
[23:00] <zwzsg> So from the look on my model I can guess where it is, and how I can enlarge my model with enlarging the collision sphere?
[23:02] <Isokron> i dont think you can enlarge the model without enlarging collision sphere at the moment
[23:02] <Isokron> but units doesnt collide between their collision spheres anymore
[23:02] <Isokron> just between their footprints
[23:03] <Gnomre> what about vertically?
[23:03] <zwzsg> Ahhhh???? Interesting.....
[23:03] <Gnomre> a low flying aircraft could fly through ground units?
[23:03] <Isokron> well air units still collide with collision spheres



You are wrong and I am right.
I have tested ingame and I refute your claim that you have.
I come with a proof:


Testing ingame > all.


PS: Screaming in bold red size 24 is too funny, I wish I hadn't started because now I can't quit it! Sorry if that post and rude and if you already agreed that collision is footprint based save for weapons and flying units which use sphere, but I can't resist the call of bold red definitive sentences about right and wrong!

[Guessmyname]

zwzsg, user of the TAU smilies (I hope you're not sucking away bandwidth), you do weapon collisions use te footprints?

[jcnossen]

Ok if SJ says that then I suppose it does use footprints
After all he actually wrote the code for it...

[zwzsg]

[jcnossen]

But that's a feature...

[AF]

That's what i was talking about with weapons collisions, the explosions are on the collision sphere.

[zwzsg]

And once again, I prove the disbelievers they are wrong:



I extracted the armpw.3do, and added a new piece at 50,50,50 to enlarge greatly the peewee collision sphere while keeping its small footprint.

[Tim Blokdijk]

I would like to see somebody update the wiki with the full story, once you all agree about collision spheres, footprints and how it works if x is... etc.

[AF]

That is not an accurate representation. A sphere is not a 2D object so seeing explosions within the sphere is not an accurate proof. See below illustration as to why that prooves nothing for your cause zwzsg. I agree that collision detection is now primarily based on footprints and not collision spheres, btut here are many different routines within the spring engine to deal with collision detection of many different types. And weapons collisions are still sphere based and not footprint based. And if they arent sphere absed then they must be box based, because a footprint is a 2D object not a 3D object and thus if footprints where used explosions would only occur on the ground around the untis feet and weapons would pass through units, or you could detonate a mine by firing at it's position but 100 click higher up in the air.....

[zwzsg]

Nice point, indeed the sphere was placed like on your drawing. Apparently the sphere is centered on the unit center for x and z, but not on y (how could it be anway, there's isn't any footprintY tag). So I remade a new armpw.3do with this time a bipoint in the new piece, from 0,-50,0 to 0,-50,0, so the sphere is welll centered even on the Y axis, and reran the test:




As you can see, the footprint is still what is used for collision on the xz plane. As for y, well, I don't know for sure. I ran an experiment where I upped the peewee health and gave the enemy some bombers, but the results are rather hard to interpret:

[Torrasque]

I'm not sure to understand everytings..is it so hard ?

And if they arent sphere absed then they must be box based, because a footprint is a 2D object not a 3D object and thus if footprints where used explosions would only occur on the ground around the untis feet and weapons would pass through units, or you could detonate a mine by firing at it's position but 100 click higher up in the air.....

Ihmo, it's bullshit


What spring do : check if it's in collision with the sphere, if yes -> it check if the missile on y,z is on the footprint, if yes -> collision.

[Masse]

OOOOOOOH... that clears it all... thanks Torrasque

[FizWizz]

what a debate! is Torrasque right? Has Zszwg's Big Red Fonttm vanquished all opposition? Or has Alantai's gratuitous consumption of bandwidth payed off and illuminated the real truth? Stay tuned for the latest in Maps: Im not dead...
maybe this thread ought to be moved to Development instead

[Zoombie]

However weapon collisions are probably sphere...

I'm fairlysure that they are.

[Masse]

well Torrasque is right on this one... u can see it from the creenshots very well if that what he said aint clear enought i could draw it

[AF]

I'm not sure to understand everytings..is it so hard ?

And if they arent sphere absed then they must be box based, because a footprint is a 2D object not a 3D object and thus if footprints where used explosions would only occur on the ground around the untis feet and weapons would pass through units, or you could detonate a mine by firing at it's position but 100 click higher up in the air.....

Ihmo, it's bullshit


What spring do : check if it's in collision with the sphere, if yes -> it check if the missile on y,z is on the footprint, if yes -> collision.

ahem...

or you could detonate a mine by firing at it's position but 100 click higher up in the air.....

Which is what is seen in zwzsg but it seems there is a constraining value for the y axis, it just doesnt seem all to consistent to be either the unti height or the sphere. Otherwise what i said remains true, you cant detect collisions in a 3D plane against a 3D object using a 2D method and get accurate results without constraining it in the y axis, creating a box in essence. Otherwise you'd get an unending box that stretched up and down to infinity which is exactly the same as:

What spring do : check if it's in collision with the sphere, if yes -> it check if the missile on y,z is on the footprint, if yes -> collision.

Which sounds a bit odd since it'd be easier to do the following:

if projectile is past NW corner of footprint but not past SE corner of footprint and y is smaller than unitheight then collision == true.

[aGorm]

Sikly green font....

I HATE YOU zwzsg!!!

Ofcourse, I dont think taht atall, I cant belive that teh lead programer is wrong and ur right yats all. However Irafuable proff you show bringsme to teh conclsuion that I should have another bash at it tonight and see if I mucked up somewere else...

ALL HAIL zwzsg

In an even larger Blue font!

Dam it, I have work tomorrow, holidays are over, and I hav'nt even got a map out.... what is teh world coming to? I guess I should use some BIG RED LETTERS now, just to make people read this sentance....

Oh, BTW happy 2006 guys...

aGorm

[SwiftSpear]

Shouldn't units target the center of a unit's collision sphere? This can still be problematic as units that are really wide will technically be really tall as well. Units with wide footprints that are really short and flat will get hit quite easily even when thier firing platforms are very low to the ground. Tall units have a distinct advantage simply because thier hit detection is acctually accurate.