Newton 1.5 released

[Sascha Willems]

Good things take their time, but finally version 1.5 of the Newton Game Dynamics SDK ('http://www.newtondynamics.com') (formerlly referred to as 1.35) has been released. There have been many changes so make shure to grab it. Some of the more interesting changes :
<ul>
Continous collision
Totally overhauled vehicle joint
Closest point calculation for convex geometry
Overhauled collision system
[/list]
And much more, including fixes speed-ups and some other smaller additions. Right now there are only SDKs for Windows and MacOSX, but Linux should follow soon.

I also uploaded the translated Delphi headers, and this time the download also contains the translation of the custom joints from the Newton SDK. Those are joints that were recreated using the custom joint interface of Newton. This is a very unique feature that allows the user to create any kind of joint he needs, instead of limiting the user to a given set of joints (like most other physics SDK do).

A big thanks goes to Jon Walton (aka Tux) for sending me the initial translation of that custom joint unit which the one included in the download is mainly based on.

You can get the download here (~20 KBytes) ('http://newton.delphigl.de/newtonimport.zip'). It includes the header and the translated custom joints and should work with Delphi and FPC.
(Note : I haven't had time to update my newton page, but I'll do as soon as I get the time for it)

If you find any bugs, please report them here or over at the Newton forums.

[technomage]

Excellent :D

Does anyone know how Newton performs against ODE? I'm currently using ODE for my project which is OK, but if Newton is faster then I might consider changing?

[Sascha Willems]

Well, since Newton has a much more accurate (and physically correct) solver it's mostly slower than ODE. But that's only in extreme cases, usually you won't notice a difference. So in the end it's features against speed, and ODE btw. isn't developed much anymore.

[Firlefanz]

Is it free to use, is it easy to handle?
Is there also a collision detection usable for an action game?

Most certain I'll do a 3D project next, and collsion detection will be
the hardest part, cause I never did something like that.
And some physics would be nice, if easy to use.

Thanks,
Firle

[Sascha Willems]

Yes it's free, no matter if for freeware, shareware or commercial software. The only thing the license forces you is to inform the makers of NGD about your project using it.

And yes, Newton also has it's fully separate collision system which will let you use Newton for collision. detection only by passing your own data. Some people already do that, though most people use Newton for all the stuff.

[WILL]

I was wondering when this was gonna come out. :)

So are the Linux and MacOS X versions gonna be ports of this same ralease or will they be released with differring features?

[Sascha Willems]

Nope, same features as Win32. Mac was released at the same time, only the Linux release isn't available right now but should follow soon.

[WILL]

Ok, excellent. Hopefully it'll be available before the start of the competition for those that want to use it. ;)

[technomage]

Well, since Newton has a much more accurate (and physically correct) solver it's mostly slower than ODE. But that's only in extreme cases, usually you won't notice a difference. So in the end it's features against speed, and ODE btw. isn't developed much anymore.

Mmmm interesting, The number of objects I'm talking about are 50-100 bodies and associated geoms for collision this will increase as I will need to add bodies and geoms for bullets and missiles etc, so I would probably be talking 500 odd objects. Is this a reaonable amount of objects? What is the maximum number you have tested before you notice a large slowdown in the calculations?

Thanks for the tip on ODE, the fact that it's not developed much anymore is a bit factor, pity all my code used ode though. I knew I should have abstracted my physics system out :wink:

[Sascha Willems]

It's not the total number of objects that matters, it's how many of them actually collide at a time. There are users (see GangWar on GameDev) that use ten thousands of objects in the same level. As long as you do some smart room divisions to keep the number of simultaneously low you shouldn't get into troubles.

[BlueCat]

This is great news :D

And thanks for getting the Pascal headers done so quickly Sascha :thumbup:

This is a good excuse to get my old engine out :shock: and have a play with Newton. I've always used ODE but Newton is looking a lot more fun these days :P



Thanks for the tip on ODE, the fact that it's not developed much anymore is a bit factor, pity all my code used ode though. I knew I should have abstracted my physics system out :wink:

I've done the same with my code :lol: I'll never learn :roll:

[savage]

Great work on getting 1.5 out the door!

Will all the current demos work with it out of the box? Will there be some new SDL/Newton demos showing off the new features?

[Sascha Willems]

Na, some functions have been removed and some renamed, so there need to be some changes made to the current SDL Demos. I'll take care of that within the next days and release updated versions of those demos.

As for demos for new features : Just wait for the release of the NewtonPlayGround. That's my main priority right now, and I hope to get it out by the end of this month.

[LP]

... version 1.5 of the Newton Game Dynamics SDK (formerly referred to as 1.35) has been released.
Do you mind if I post this on Turbo as well? :)

[Firlefanz]

On the german DelphiDev.de site they often post the news from PGD, and they always tell where they got it from and link the PGD main site. :wink:

Firle

[Sascha Willems]

Do you mind if I post this on Turbo as well? :)
Certainly not ;) The more widespread my news get the more comfortable I feel, so feel free to use all the news I post here also on turbo.

[Sascha Willems]

Attention to all users of the custom joint unit :
I was recently e-mailed about errors in the custom joint unit. By accident I included an older version of that unit which was using the old name of a function from the Newton beta SDK and therefore couldn't be compiled. So please re-download the package as I uploaded a fixed version of that custom joint unit.