Exact solver disables joints

Report any bugs here and we'll post fixes

Moderators: Sascha Willems, Thomas

Exact solver disables joints

Postby pHySiQuE » Thu Sep 05, 2013 3:21 am

It appears hinge, ball, and slider custom joints are disabled now when the exact solver (0) is used. Those are the only joints I currently support, so I don't know about others.
pHySiQuE
 
Posts: 608
Joined: Fri Sep 02, 2011 9:54 pm

Re: Exact solver disables joints

Postby Julio Jerez » Thu Sep 05, 2013 8:38 am

I have not work much with the exact solver. what are you using it for?
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Exact solver disables joints

Postby pHySiQuE » Thu Sep 05, 2013 12:44 pm

The exact solver is the default setting in my engine.
pHySiQuE
 
Posts: 608
Joined: Fri Sep 02, 2011 9:54 pm

Re: Exact solver disables joints

Postby Julio Jerez » Thu Sep 05, 2013 12:49 pm

that makes everything slower. why don yo umake mode 1 teh defualt?
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Exact solver disables joints

Postby pHySiQuE » Thu Sep 05, 2013 1:24 pm

If it isn't supported, you should just say "this is no longer supported". Having features broken and in limbo is much worse than not having support for them at all.
pHySiQuE
 
Posts: 608
Joined: Fri Sep 02, 2011 9:54 pm

Re: Exact solver disables joints

Postby Julio Jerez » Thu Sep 05, 2013 1:34 pm

it is supported, what I am saying if that a game engine should not be using that solver.
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Exact solver disables joints

Postby zak » Tue Nov 12, 2013 2:48 pm

I have converted my project from 2.35 to 3.11 but i have a problem with exact solver.
In debug mode when i call NewtonUpdate i obtain an assertion in dgWorldDynamicUpdate::CalculateForcesSimulationMode
(In release mode joints simply don't work)
This happen only with exact solver.

Can you help me, thanks.
zak
 
Posts: 87
Joined: Mon Dec 06, 2004 9:30 am

Re: Exact solver disables joints

Postby Julio Jerez » Wed Nov 13, 2013 11:37 am

has you try using the iterative solver?
can you tell me what you are doing that need a better solver?
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Exact solver disables joints

Postby zak » Wed Nov 13, 2013 12:40 pm

A sporting game with ragdolls developed with custom joints.
With 2.35 i have obtained the needed precision only with exact solver.
I have obtained more than enough performance too.
So i have no reason to pass to iterative.

I think that for some situations exact solver remains the best solution,
but i am afraid, judging by your answer, that you no longer want to support it.
In the precedent post you assured that exact solver is still supported.
So why is it no longer working?
zak
 
Posts: 87
Joined: Mon Dec 06, 2004 9:30 am

Re: Exact solver disables joints

Postby Julio Jerez » Wed Nov 13, 2013 2:40 pm

The exact solve is working. I commnet out because I was adding and optimization tha may cause it to break.
I see fi I can re enable it.

however if you want it for rag doll, nwo ta yo uare in netwon 3.xx there is solution that yo ucna use for teh rag doll.
Netwon 3 has a newt funtion for soem kind of Artcualtions.

if and contation cna be expressed as a to down tree, (stuff like vehicles, and characters, rag dolls)
then it is possible to correct the joint error after each integration step. teh syetem is called CustomArcticulatedTransformManager

if you take a look at the ArticulatedJoints and Ragdoll you will see that is use that feature
those demo use very high mass ratios, and teh behvieo is very accurate, I think but is use the iterative solver.

The reaosn why tshi is acurate is because on each integration step the solve will fing an first order aproximation o fteh solution.
whoever because each joint have a specified gemetry. Thsi make is possible to proeject the erro ove teh geometery of the joint.
then on the next integration step does not integrate the error that was accumulater by the preview step since that error was truncated.

anyway I see if I re enabel the exact solver.
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Exact solver disables joints

Postby zak » Thu Nov 14, 2013 2:09 pm

Julio Jerez wrote:The exact solver is working. I comment out because I was adding and optimization tha may cause it to break.
I see fi I can re enable it.


Great! A new optimization on exact solver. The best physics engine, as Newton is, must have the best exact solver! :twisted:
zak
 
Posts: 87
Joined: Mon Dec 06, 2004 9:30 am

Re: Exact solver disables joints

Postby zak » Sun Dec 22, 2013 1:23 pm

Hi,
waiting for you to re-enable exact solver i worked on porting my graphics engine and gui from directX9 to directX11 and directWrite.
Now it's time for me to return on core part of my soccer game.

There are chances of having exact solver in Newton 3 for the end of 2013?
I'd not want to return on Newton 2.

Thank you.
zak
 
Posts: 87
Joined: Mon Dec 06, 2004 9:30 am


Return to Bugs and Fixes

Who is online

Users browsing this forum: No registered users and 9 guests

cron