Report any bugs here and we'll post fixes
Moderators: Sascha Willems, Thomas
by Julio Jerez » Sat Apr 28, 2012 3:17 pm
I see, that exposses quite clear.
I will debug it tonight.
It is is good you used the serialization, it makes so much easier to share information and debug stuff this way.
most important the bug can be reproduce each time.
Now this bug is personal between me and him, I have to get that scene working, that bug is not going to get the best of me..
-
Julio Jerez
- Moderator
-
- Posts: 12249
- Joined: Sun Sep 14, 2003 2:18 pm
- Location: Los Angeles
-
by Julio Jerez » Sun Apr 29, 2012 8:03 am
ok I just run this in debug mode. and I see what is going on.
The high energu bug is nor there any more, what you are seeeing in collsion malafunstion because of deep penetration and shape pieaking paralle face of the wall of the jar.
this generate contact in oppositedirection that will should that.
Thsi is in fac a very bad problem but it si no a bug, in is the diecret collision do not handle that.
when I reenable contrinue collsion then those case will be detected before they occur an teh will be condiniled properlly su that the code do no genertad contact facing away for each other.
The work for contineu collsion is amost completed I simpel nee to re eneabel all of teh inetface functions
here is a descrtion of the bug.
say you drop just one penny form a distance sufucentlly high that the frame before the fore collsion teh velocity is high enought.
that when it collide in penetrates the wall direclly facing the penny, but not the wall paralle to that interion face but face the outside of the jar.
a decrete collsion system will cacluete teh cantact tah best represent the face and teh object. but we know that the contact normal can only face to the outside.
with out any extar information ther si no way to know if teh contact is a good contact or a bad contact.
the concat is send to teh solver an dteh resposn is to move teh penney away at even high speed than is already has.
even wose tham that is tha tteh peney intesect both faces, now is is umpredicatble wh ateh solve will do, simpel because the is an unsolvable problem.
the continous collision system use the velocity of the penny to discriminate the faces in the next timestep that face away from the trajectory and filer those out, sending only the ones that oppse the mostion.
it also decimate teh time step to anvanc eteh simulation to the first time of contact so that is doe no generates contacts outside the boundaiys of neither shape.
it does that until either it reach a max step count or the complete step is resolved.
the reason CC is not enable yet is because CC is a layer on top of teh Decrete collsion system. and I had to complete all of the collision instance and compound collsion using instance before I moev to CC.
I hav etha alreay so I can resume the CC
-
Julio Jerez
- Moderator
-
- Posts: 12249
- Joined: Sun Sep 14, 2003 2:18 pm
- Location: Los Angeles
-
by Bird » Sun Apr 29, 2012 10:33 am
the reason CC is not enable yet is because CC is a layer on top of teh Decrete collsion system. and I had to complete all of the collision instance and compound collsion using instance before I moev to CC. I hav etha alreay so I can resume the CC
Glad you know what it is and can fix it.
Serialization is definitely a great addition to Newton!
-Bird
-
Bird
-
- Posts: 623
- Joined: Tue Nov 22, 2011 1:27 am
Return to Bugs and Fixes
Who is online
Users browsing this forum: No registered users and 5 guests