What's next for Newton?

A place to discuss everything related to Newton Dynamics.

Moderators: Sascha Willems, walaber

What's next for Newton?

Postby AntonSynytsia » Mon Mar 20, 2017 9:33 am

The new joint system in the latest 3.14 is amazing! Very robust, stable, and realistic. In previous versions, when multiple bodies connected with joints would spin, occasionally, there would be an external force damping the rotation. But now, with the new joint system, there are no loger rotation damps and the bodies behave as ecpected.

Furthermore, I thought l would have to rely on skeleton containers... But nope, all that is now done automatically, which simplifies things for the end user, yet presumably is very complex at the core. Even the API for skeleton containers was removed presumably do to it being no longer necessary.

The function for controlling joint solver is also a very desirable feature. I don't quite understand when to use a solver model of 1, but solver models of 2 and 0 are explanatory.

I wonder why none of these fearutes were listed in the news. I saw progress on github, but have not noticed highlights on that.

For quite a while I was relying on an outdated version, back when exact solver model was available, but now I'm very satisfied upgrading to the latest one.

Those are very useful features to have and Julio did excellent work on them.

I wonder what's next for the physics engine. Would it be soft bodies and cloth?
AntonSynytsia
 
Posts: 193
Joined: Sat Dec 28, 2013 6:36 pm

Re: What's next for Newton?

Postby Julio Jerez » Mon Mar 20, 2017 4:13 pm

Oh I am glad you like it, yes I went back to our mission statement
Newton Dynamics implements a deterministic solver, which is not based on traditional LCP or iterative methods, but possesses the stability and speed of both respectively. This feature makes Newton Dynamics a tool not only for games, but also for any real-time physics simulation.


That I abandoned for a while with early versions of 3.xx, then I tried to put the responsibility on the end user with skeleton interface, but finally I decided it was too complex for any user and decided to make transparent to them.

This is the same theoretical solver we had in 2.xx and lower but it has been refined to deal only with bidirectional joints at first then improved to deal with all kind of joints except contacts.
Soft joints was something that bothered me a lot and I had to address.

On the solve modes 0, 1, 2
mode 0, means the code will use that as a hard joint whenever possible. this is good for almost any kind of articulated model.
mode 2, mean the joint will use the iteration solve every time, this is good for joint that are issue in real time and can form estrange configuration, stuff like picking joints, dry frictions, breakable joints, etc

mode 1 is the most interesting this is for when the user want the joint to be hard, but the joint can form a kinetic loop with other joint and decision has to be made. For example say you have you have a human ragdoll with a two handed weapon.
The arm, the body and the weapon form a closed loop so one joint need to be part of the generic LCP solver, if we let the algorithm to pick which the joint, it may pick a bad candidate like the arm joint,
the user in these cases would know that the point when the weapon attach to the arm is a better candidate, so he can tell the SDK that that joint is the one to blame.
Basically this option is a tide breaker decisions. the joint are all solved exactly it is just the one is selected for the slower solver path.
In most cases you do not have to worry about because the custom joints already made that decision.

As to what next, yes we are going to address the soft bodies and cloth, we will probably leave for a new release.
immediately I will try to elaborate more on the vehicle module, tanks, for wheel drive, and other things. and we will close 3.15 with at least the beginning of a self balancing character.

If you have any particular feature you would like to see please tell me.
I always give priority to user demanded feature specially of the do put the to good use.
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: What's next for Newton?

Postby AntonSynytsia » Tue Mar 21, 2017 1:01 am

Awesome, I like the current schedule!
AntonSynytsia
 
Posts: 193
Joined: Sat Dec 28, 2013 6:36 pm


Return to General Discussion

Who is online

Users browsing this forum: No registered users and 12 guests