Piece and mesh collisions - Page 2

Piece and mesh collisions

Requests for features in the spring code.

Moderator: Moderators

User avatar
Argh
Posts: 10920
Joined: 21 Feb 2005, 03:38

Re: Piece and mesh collisions

Post by Argh »

I must be missing something. If footprints must be square then why are they defined using footprintX *and* footprintZ?
Because that's used for some visuals. Footprints were square in OTA for mobile things too, IIRC. Immobile things use Yardmap. I think that the ultimate solution is to allow mobile Units to have a Lua-manipulatable Yardmap, and manage it via Lua entirely, to cover special cases.

But that's just me, and I'm just theorizing out loud. I'm pretty sure that's the way to go, though.
Kloot
Spring Developer
Posts: 1867
Joined: 08 Oct 2006, 16:58

Re: Piece and mesh collisions

Post by Kloot »

FLOZi wrote:
the old offset tag only affected 3DO's in the first place
Assuming this refers to my offset tag, that is incorrect.
You don't remember your own code very well, then, because it didn't touch S3O's in 0.75b1 when it was introduced, nor in 0.76b1+svn when it was replaced.
You missed all of smoth's being upset
Yes, shockingly.
at the fact that his flat units in gundam, most notably air fields, have units aim completely over them, doing no damage? They're 3do, so he can't move the 'center'. From what I've seen these tags get a good bit of use on 3do units, and any ability to set an offset would be very nice.[
Looked into this, I'll admit it's problematic for zeonairport and zeonengineering. I can add the tag back in, but the easiest way for smoth to deal with it is to just make the volumes bigger along their y-dimensions. Those two buildings are only half-covered in 1.2f2, and not because of their negative vertical offsets:

download/file.php?mode=view&id=1667 (with offset)

download/file.php?mode=view&id=1666 (without offset)
It's troublesome when spring doesn't even show the aim center that units use.
As I said, the collision volume center (before offset is applied) is the aim center.
Another mantis ticket is that units don't use the collision shapes for aiming, but the invisible model sphere, so units think they can't aim past allies when they easily can
That's a to-do item.

OT: I may prototype a piece-based coldet system, but meshes are IMO a bridge too far for an RTS engine.
Attachments
withoutOffset.jpg
(144.99 KiB) Downloaded 31 times
withOffset.jpg
(127.63 KiB) Downloaded 30 times
Last edited by Kloot on 15 Apr 2009, 13:41, edited 1 time in total.
imbaczek
Posts: 3629
Joined: 22 Aug 2006, 16:19

Re: Piece and mesh collisions

Post by imbaczek »

convex hull might work, but it's overkill in my opinion, too.
User avatar
SpliFF
Posts: 1224
Joined: 28 Jul 2008, 06:51

Re: Piece and mesh collisions

Post by SpliFF »

OT: I may prototype a piece-based coldet system, but meshes are IMO a bridge too far for an RTS engine.
That would do all I actually need and should provide a reasonable level of accuracy if I'm careful with my modelling.

I would recommend the addition of a unitdef tag to go with this:

noCollidePieces = {'muzzle','smoke_point','antenna'}

To ignore small or internal parts that would just waste cycles. The code should also ignore pieces hidden by COB or Lua scripting.
User avatar
smoth
Posts: 22309
Joined: 13 Jan 2005, 00:46

Re: Piece and mesh collisions

Post by smoth »

Kloot wrote:
You missed all of smoth's being upset
Yes, shockingly.
you saying I need to rage louder? Is that endorsement for raeg?
Kloot wrote:Looked into this, I'll admit it's problematic for zeonairport and zeonengineering. I can add the tag back in, but the easiest way for smoth to deal with it is to just make the volumes bigger along their y-dimensions. Those two buildings are only half-covered in 1.2f2, and not because of their negative vertical offsets:

download/file.php?mode=view&id=1667 (with offset)

download/file.php?mode=view&id=1666 (without offset)
yes but that is really bad as they are shooting in the air above the airport, not at it. The experimental was because when I play in fps it would block clicks covering the whole thing. Are you looking at fixing/bringing back the tag?
User avatar
lurker
Posts: 3842
Joined: 08 Jan 2007, 06:13

Re: Piece and mesh collisions

Post by lurker »

Kloot wrote:
You missed all of smoth's being upset
Yes, shockingly.
Well he does go off in #sy a bit..
Kloot wrote:the easiest way for smoth to deal with it is to just make the volumes bigger along their y-dimensions. Those two buildings are only half-covered in 1.2f2, and not because of their negative vertical offsets:
The control tower on zeonairport isn't that big, and as your image shows the hitbox would have to cover vast amounts of empty space to stretch from the ground up to enclose the aim point, making it impossible to fire over the unit, even from a high position, with midair collisions on nothing. Edit: And yeah, the units would still be aiming at air, only now the air would get hit.
Kloot wrote:
It's troublesome when spring doesn't even show the aim center that units use.
As I said, the collision volume center (before offset is applied) is the aim center.
But this isn't shown in debug mode.
Kloot
Spring Developer
Posts: 1867
Joined: 08 Oct 2006, 16:58

Re: Piece and mesh collisions

Post by Kloot »

smoth wrote: you saying I need to rage louder? Is that endorsement for raeg?
So long as you raeg in PM. ;)
yes but that is really bad as they are shooting in the air above the airport, not at it. The experimental was because when I play in fps it would block clicks covering the whole thing. Are you looking at fixing/bringing back the tag?
Yes, it's restored (but renamed to modelCenterOffset). And now it does work on S3O's.
The control tower on zeonairport isn't that big, and as your image shows the hitbox would have to cover vast amounts of empty space to stretch from the ground up to enclose the aim point, making it impossible to fire over the unit, even from a high position, with midair collisions on nothing.
Either way, it's an approximation.
But this isn't shown in debug mode.
Fixed.
User avatar
smoth
Posts: 22309
Joined: 13 Jan 2005, 00:46

Re: Piece and mesh collisions

Post by smoth »

Kloot wrote: Yes, it's restored (but renamed to modelCenterOffset). And now it does work on S3O's.
That will be in the next spring?
Post Reply

Return to “Feature Requests”