Custom joints vehicle and weak joints

[Julio Jerez]

That paper is does not really works.
It really is a very naive implementation of Featherstone, that present to overcome the serious mathematical limitations of Kinematic loops by simple doing Iteration.
In reality when you try to solve kinematic loops that way, you enter into a realm of very serious Stability problems. This is some of the Battle if have with the so call industry experts.

Anyway, I believe I do have a solution, but I have not oput to much wrk into it.

Basically I will use Featherstone formulation post process to run and Post possess pass to correct kinematic drift father the solve ass.

Basically I will let the General Lagragian solver find the solution, and the use a Generalize Coordinate solver to refine the solution.

There is a very crude implementation of the in the Moltibody car solver, by using simpel projection,
but if have some bug that ill formulations.

[kallaspriit]

I hope you'll find some time to look into this, stiffer joints would benefit many NGD users and especially applications like my complex suspension joint-based car simulator

[PJani]

Any news about stiffer joints in newton?

[Julio Jerez]

Pjani, I know it has being a long time since I said I was going to work on that. Here is a rundown of where I am.

Newton is now at a point where any new feature requires a framework to work with, each time I am going to work on something I have to spend lot of time setting up the example,
and when I have something going I always face the problem that the end user find it too complex, or cannot use it because of the Interface.
Or it is just false start and I have to start all over again, even worse if I have to stop and resume later I find that I have to learn all over again the same demo,
and usually I end up rewriting a new one.
On top of that every time I have to help someone I have to spend lot of time studding their style, and some people use Newton in library format, so it is even harder to help them.

That was Ok when the engine was small, but as the engine get more complex it is becoming too difficult for me to handle.
so I decided that the way around this is to have a Graphics Visual interface where all the features the engine have to offer, can be put together graphically.
In other word a Modeler for content creation.

At first thought of buying 3dsMax, but most Newton users cannot afford max anyway, a while Back I bought XSI but soft image left me stranded with an buggy modeler with unsupported SDK.
I also took a look a Blender but it is prohibited to extend Blender unless you write in Python plus blender's horrendous interface is quite mediocre.

So I decided the make my own editor with a look similar to Max but not quite exactly.
you can see the state that it is at, by downloading 2.24 (change the name on the download)

when it is in a state that is usable I will recreate all of the Netwons demos in the SDK using the editor and then the end user they can export then to a Database,
kind of like a blueprint of the Physics and Graphics scene.

in the end application can have any choices.
-link to the NewtonArchemendia Library and write databases.
-write a plugging to interface with editor and export and imports to their own format,
-Or even write a wiewer window plug in tah can let the cliend application see the seen in real time.
-or Parse the xml or Binary format exported by the editor.

the editor is a Plug in architecture, the end use will be ably to write Plugins and make it their own editor if they like.
After I get the object Navigation working I will integrate the Physics modeling with all of the featyre supported by Newton.

It is at that time that I will address all those idea I have for stiff joints and oteh features but that I am unable to implement because of the lack of framework.
thonkg like perfect the Player controller, Rigid Joints, Perfect the Raycast car, SoftBodies, Destruction with debris, Animation adaptation and retargeting, and so on.
all require a visual fremeword because there too complex and art driven to be provboded as a set of functions in a library only.

This will be more than just an editor, it will also have modeling tool with limited graphics capabilities, like low polygon editing, Solid modeling brush style,
spline/Nurbs, special models like 2d and 3d terrains, Skinning, animations by inverse Kinematic and constraints.
I am hoping that with an user friendly Editor with the campability to export all of the Data to any graphics engine.
we can move from away the procedural way of modeling to a Visual kind of "what you see is what you get style" of Physics and Graphics modelling with newton,
and with the runtime library will make eassy to get the result into another applications.
something like

// load a complete Physics workd.
Database->LoadScene (newtonWord);

// use a binding to look up for teh objecst teh application is interets in
//Parce Phsyics world
DateBase->IteratePhysics()

//Parce Physics world for joints and Controller
DateBase->IterateJointss()

// Load Graphcs
DataBase->IterateGraphics();

// Load Animations
DataBase->IterateAnimationss();

...

Maybe then more Artisth oriented people can take a look and the engine and generate better quality work with it.

So that is what is holding me up on the more solid joint promise, it is no until I get this over with that I will add more features to the engine,
because with each new feature is a new set of interface problem.
Yo can see that pattem in the SDK demo. basically each demo is a complete new application, and that can not scale up eassilly.

[PJani]

Thanx i hope i didnt took to much of your time and/or bother you .
I only wanted to know how stuff goes.
You rulz with your newton engine, keep the good work as you did before.

[kallaspriit]

Great to hear that you have such big plans for NGD

Hope you will someday find some time to try to create some really rigid hinge joints, the main thing I am missing in NGD right now