|
|
Physics, objects pass through objects
About Truespace Archives
These pages are a copy of the official truespace forums prior to their removal somewhere around 2011.
They are retained here for archive purposes only.
Physics, objects pass through objects // Tech Forum
Post by Finis // Feb 1, 2009, 7:42pm
|
Finis
Total Posts: 386
|
Model side.
I'm trying make a wall of blocks to knock over using the physics simulation. The blocks sink into to the ground. They start a little above the ground and there is a 0.01 inch space between each block.
The ground is a box, not a plane. I have quad divided the area under the blocks into small rectangles and set the simulation properties to detect vertex and face collisions.
I had no trouble with many simple boxes with physics in my Splododyne WIP thread.
Any suggestions to keep the wall from sinking into the ground? |
Post by spacekdet // Feb 1, 2009, 9:26pm
|
spacekdet
Total Posts: 1360
|
Is that just a section on the ground cube object that's been subdivided, or the entire top face?
Maybe try subdividing the whole top face- you may not need it as dense as you have it shown in your screencap.
Did you set the properties of the ground to be 'iron'?
Right clicked on collision detection and chosen 'high quality'?
I just ran a couple tests, only I raised the 'ground' cube above the grid.
Bounced right off, no sinking in. Not sure if this makes any difference or not.
17991 |
Post by Georg // Feb 1, 2009, 9:28pm
|
Georg
Total Posts: 270
|
Can't you put the plane on the ground? (z=0). Do your blocks interact with each other? if so,, it must have something to do with the number of vertices in your blocks. I see it often with physics that object interact unless you quad divide them a bit more.
Georg |
Post by Finis // Feb 2, 2009, 7:44am
|
Finis
Total Posts: 386
|
Georg - Thanks. I could, and might have to use the ground plane as the ground. That would work here but I'm also trying to learn about the abilities of the physics system. What if I wanted items on top of different height blocks? Only one could be at the ground plane. What about a lumpy terrain with objects resting on different places?
Yes the objects will be highly interactive. This is for my Splododyne Explodiatory Explodolator wip. If I have to subdivide hundreds of blocks then the simulation would take forever to run.
spacekdet - Congratulations on the MMC win. I set the collision to face and vertex. I didn't know the other collision thing, where the precision thing is, was for physics. Checked it and it was on high precision. Only the polys under the wall are subdivided.
I ran a huge test: my Splododyne wip. The two tests there so far (this is to be number three) had only simple boxes and I had no trouble there. There was no quad divide there just six poly boxes. The boxes didn't have much time to sink though.
Iron floor? Hmmm. The floor has no physical properties. I'll try giving it properties and putting the bottom on the ground plane so it won't sink.
Can the object fixation points be used to make an object with physics properties immobile? Has properties for friction, elasticity, etc. but won't move? |
Post by Finis // Feb 2, 2009, 11:39am
|
Finis
Total Posts: 386
|
Ground pass through solved.
Model side.
Spacekdet recommended that the ground have physical properties. I hadn't assigned it any. I gave it properties and then the blocks didn't sink in anymore. Even when the ground surface was one un-quad-divided face of a box. It seemed to run the simulation faster too.
Then I discovered that there is no documentation for model side physics! The adhesion property looked promising to keep the block wall from disintegrating completely. I experimented with adhesion and found that things moving at low velocity would stick together on first contact or behave as if there was no adhesion when the setting was changed the least amount allowed. Adhesion of 10-20 would produce reasonable results for high speed collisions and together with friction seems to help stop the objects from sliding or vibrating forever.
Objects still have strange behaviors. The attached movie shows that it is possible to partially destroy a wall. Note that some odd sliding occurs and one block is unrealistically suspended by adhesion no doubt. Rebar around the top of a wall might hold a block like that but there is none modeled. More friction might fix the sliding.
Sometimes an object will stop moving but then move a little more after a short time.
The movie shows an 18kg block fired at 100m/s into the wall. The blocks are simple beveled boxes and the ground box is not divided on top. |
Post by Nez // Feb 3, 2009, 2:10am
|
Nez
Total Posts: 1102
|
Then I discovered that there is no documentation for model side physics!
Yes, I thought I couldn't find anything... I still have tS5 and the pdf manual for that, so I could post/email you the relevant chapter (ch.21) from that manual if that would help - I would imagine that the tS5 physics are very similar to the 6.6/modelside physics although there might be some differences.... or you cuold try an admitted 6.6 user (e.g Weevil, Transient) to see if they have a pdf manual from 6.6 where they could send you the relevant chapter... |
|
