How NGD works

[kallaspriit]

Great, hope you can find a solution to this problem.

Have you thought about the questions in the first entry? My paper deadline is approaching fast and I could really use this information to be more specific when introducing NGD. I'd really appreciate if you could find some time to shed light to the inner-workings or this great physics engine

[kallaspriit]

I have updated the demo with some new handling, swaybars, suspension geometry and visuals

The new shell:


Rear suspension geometry:


Help is displayed in-game, but the most important buttons:
F1 - toggles the in-game help
F4 - toggles debug-arrows display
F5 - toggles physics pause
F6 - toggles free roam and car follow camera
F7 - toggles slow-motion mode, great for seeing how suspension works under different circumstances
Hold CTRL to pick items
Hold SHIFT to use twice the torque

Use vsync for fluid follow-camera.

Download

I am still hoping for answers to the questions in the first topic, my paper is getting ready and deadline approaching

[Stucuk]

You have too much specular lighting on objects. Cones shouldn't blind you when you look at them (Plastic cones are generally dull, while the White bits are reflective to light, they don't shine that much)

[kallaspriit]

Thanks for the tip, you are of course right, I haven't put much effort into the visuals - just slammed on a texture and that was it, will fix it someday.

Notice that the car has no drive-drain yet (engine, gearbox, diffs). If anyone gets this to work on their computer, any feedback is welcome

[kallaspriit]

I have updated the demo a bit (mostly behind the covers) and fixed a problem that caused crashes, the new version is downloadable here.

Julio, have you tried the new demo? Could you please find some time to give some insight on how NGD works so I could write a paragraph or two about it in my bachelor of science degree paper, that I need to have ready by the end of May.

1. What methods are you using
- general
- solver
- collision detection
- friction
- ...

2. Why are you using the methods you are
- accuracy vs speed vs ?
- what is good and bad with these methods, what are the limits
- what would be other possible methods, why are you not using these
- how accurate is the engine at simulating real world rigid body dynamics (vs other physics engines)
- what is different about NGD and other engines (why is is better, more exact)

3. How joints work
- how are they implemented
- what are the limitations (for example maximum mass-ratio)

4. Why aren't NGD joints in current implementation stiff enough to simulate the kind of arrangement I wanted to use for my vehicle
- what could solve this problem (you mentioned Fetherstone solver)
- how does Fetherstone solver solve the problem better then currect implementation, what are the limits of using it

Any info would help, I would really appreciate it

[Julio Jerez]

Oh I am sorry I miss it.

I appologioze, I keep falting on you.

let us do this.
I will write the answer to you this weeknd and send them to you by PM.

I have done this for some othe people hwo in teh pass need soem more thatrical frame abput Netwon.

I do no like to do for people who are making demos and Games because newton in not a subtitute for Theoretical Reseach,
netwon is just a Physics library.
But some people need to have some of the Theory behind Netwon so I will reply to you.

[kallaspriit]

Great, will be looking forward to it.

My paper needs to be ready by Tuesday so getting the answers this weekend is critical to include this info

[Babbo]

Hey Julio, these questions are mainly for you and I thought I'd ask it in the forums instead of via PM, perhaps someone else is also interested and this could be added to wiki for everyone to see and read.

I am writing my bachelor of science degree paper using NGD to simulate vehicles made out of various bodies and joints, remember the Custom joints vehicle and weak joints thread.

I want to write a chapter on Newton Game Dynamics and it would really help if You could explain some aspects of NGD:

1. What methods are you using
- general
- solver
- collision detection
- friction
- ...

2. Why are you using the methods you are
- accuracy vs speed vs ?
- what is good and bad with these methods, what are the limits
- what would be other possible methods, why are you not using these
- how accurate is the engine at simulating real world rigid body dynamics (vs other physics engines)
- what is different about NGD and other engines (why is is better, more exact)

3. How joints work
- how are they implemented
- what are the limitations (for example maximum mass-ratio)

P

Hi, i'm a computer science's student and i'm writing my degree thesis. I'm using NGD to implement a dynamic simulator for autonomus wheeled robot driven by an artificial neural network. it will be very helpfull if you can send me the same answer at the kallaspriit's question. Thank you very much!
PS: if you are interested, i can post my work.

[Julio Jerez]

This is a big task you ask me.
I need to write and engine user manual, because it has become so large that even I start to forget some of the funtionality.
I was thinking to wriet a PDF documnet with index,

I make you a deal. I start writing the doc, and I can give you my notes and you act as an editor and fix my grammar and syntax.
then I used that you the user manual, and you are free to use for you thesis.
you use what you need and when you finish you do no have to continue helping.

Few people has ask for explanation and I alway do via Email and PM, but the problem I have with the that is that after few weekd of conpromise
I am always left with zero gain for my time. This time I do not want to do that privatly.

can you tell me what do you have in mind?

[Babbo]

An user manual will be very useful for every NGD user. It's a good idea but also a very challenging one.
It will be my pleasure to help you with this work. I can extract my necessary information from the notes.
What I need for my thesis is an intorduction to the NGD, explaining main design choices and algorithms used, engine's limitations. Not the specific implementations, but only the choise made and why.