|04-02-2012, 09:04 AM||#1 (permalink)|
Join Date: Dec 2007
Location: Somewhere purple, Germany
SL Join Date: 20. January 2007
[SVC-7800] 'None' physics setting getting the same punishing LI calculations for com
A little JIRA I just opened after noticing this.
Basically, objects get a horrible LI punishment if their linkset contains things like toruses with a 'prim' setting, and also contain child prims with the 'convex hull' setting.
So far so normal, but it seems said LI punishment also takes effect when the child prims are set to 'none'!
This is definitely a bug, as there should be no additional physics strain in that case at all, and the 'none' setting is actually very advantageous to the simulator.
Added an example. This will have 67 LI:
root: cube, prim
child: torus, prim
child: whatever, none
Which is expected if said 'none' is convex hull, but the 'none' setting shouldn't screw it up like that.
"Have you ever noticed that anybody driving slower than you is an idiot, and anyone going faster than you is a maniac?" - George Carlin
Last edited by Chalice Yao; 04-02-2012 at 09:12 AM.
|04-02-2012, 09:24 AM||#4 (permalink)|
Join Date: Apr 2010
Oh yes, no dispute the current system is wrong. It makes mixed content a nightmare.
Remember when it was said "Even if you can't model, you will benefit from mesh because you can use them in your builds as custom primitive shapes!"
|04-02-2012, 09:40 AM||#6 (permalink)|
You look tasty.
Beeble Bubble Bimbo and Griffer Alt but not a Dog Chewy
Join Date: Aug 2007
SL Join Date: October 24, 2006
Client: Frestorm, Imprudence
Blog Entries: 3
I showed someone that my mesh tiki-head plus a spiraled and hollowed torus could have a LI of 1536...
SL easily allows you to accidentally your entire pants. -- Adeon Writer
|1 User Agreed:|
|04-02-2012, 11:24 AM||#8 (permalink)|
Tea swilling lurker
Join Date: Oct 2009
SL Join Date: Feb 2008
Business: Tatiana's Tea Room
Voted & watching. Thanks for this. I was always shocked at the results I get when testing out "none". Makes no sense to me at all.
Fine French Antiques ♥.•:*¨¨*:• ♥ Tatiana's Tea Room ♥.•:*¨¨*:• ♥ Surround Yourself with Classic Beauty
|09-24-2012, 10:10 AM||#10 (permalink)|
Who are you?
Join Date: Nov 2010
Location: Somewhere else.
SL Join Date: Early '06 or late '05, not sure.
It makes sense to have a non-physics cost exaggerated because of physics?
I am not sure I've ever used non-physics prims in a build as it almost never does anything good and usually sends it into the trash heap.
|09-24-2012, 10:22 AM||#11 (permalink)|
Join Date: May 2009
SL Join Date: 17 June 2007
Business: Something Spunky
As I understand it, the issue is that, by setting one of the child prims to either convex hull or none, that ensures the mesh accounting system is used for the whole build.
The physics cost of a torus left as PRIM_PHYSICS_SHAPE_PRIM (in this example, the root prim) is very high. Switch it to PRIM_PHYSICS_SHAPE_CONVEX and the cost goes down.