Is newton really scalable?

[Julio Jerez]

Now I added the secund part to the Demos.
the Physic is runs in a separate Thread, that allows the physic simulation to take a Full Frame for rendering withou interfaring with the graphics.

By doing This your scene now runs from 1600 fps and drop to 1200 in debug mode. (before is was dropping to 87)
In release is it from 1200 to 2000 and stops there, I am gessingn this is the limit of the refress rate of my system because it does not goes any fater even if nothing is done.
The physics is runned at fixed 120 fps and the frame are smottly intepolated, there is no shoppined at all in the simulation.

I have to add few more basic features, but the good thing is that this is also part of the SDK and I will integrated in Ogre New this weekend.
so for people wuth Multicores and OgreNewt should run much smother after this secudn update since the teh job in the
BasicFrameListener will no longe be serialised wit teh rendering thread and all the other stuff Ogre is doing.

[Julio Jerez]

I do not kwon if you are still waching this.

I have OgreNewt runin the pheyis ina separate Thread, and all demos nwo run at teh top Ogree speed even in debug mode.
The physic is not a factor anymore.
Running in separate Thread can be Toggled bu teh applytion ate run time so teh teh diffrent in preformance can be mesuare in real time.
Running the physics asycronous is a Huge preformeace Booth when running a Multicore CPU.

Later I will add a Ogre Demo that load Alchemedia files and see how your scene performs in OgreNewt.
Unfortunally I have to work this weeked, because of tght milstone, so I will continue tomorrow.

[Slick]

Julio will you be adding the other newton demos into the Ogrenewt framework? I have been playing around with the friction movement track demo in newton and ported it to ogrenewt but I can't get it to move and rotate correctly. I think it must be some of the matrix conversions or something.

[Julio Jerez]

Yes I will add then all,
In fact all newton applications will be able to play on any Graphics engine that Links the dScene library and write the glue code to get build the graphics objects form the dScene Mesh.
I am adding the part tomorow.
Basically I will not add the demos instead Scene Saved from the demos will be loadable in any OgreNewt application.
and any Scene crate using OgerNewt and save to a dScene (physics onely or Graphics and Phsyics) will be loadable on the Newton demos.

what I really need now is that some one in the OgreNewt from integrate the Patch so tah I can post a newer Patch.

[razi]

Of course Im still watching this, I am patiently waiting for your news . Oh, but about updating ogrenewt repository, there were some errors compiling it on linux, the errors like some headers didnt exist, mostly from dPluginsStdafx.h, so they wont commit it.

[Julio Jerez]

yes there will be error in Liunux and Mac because I have no make a Mac SDK,
can the patch be made and notify people that it is only available in Windows until the SDK in released for Linux?

Must thse file are in the SDK, an dtehy are open source, anyone can fix the Linux build if they want.

[Julio Jerez]

I have a question,
I OgreNewt how do you show the debug collision?
Is there a key for that?

[razi]

F3 should work

[Slick]

Basically I will not add the demos instead Scene Saved from the demos will be loadable in any OgreNewt application.
and any Scene crate using OgerNewt and save to a dScene (physics onely or Graphics and Phsyics) will be loadable on the Newton demos.
.

Doesn't loading a scene preclude people from gaining knowledge from looking at how the scene/effect was done in code? Unless I am missing something examples are so that people can learn features of the library.

[Julio Jerez]

Not it does not, all of the file are open source.

The code is standarized form app to app, no the mess we are gutthiong now when avery open use hsi own method and some tine we teh get it right and soem time teh getti wrong.
Plaus the file can be seem in the levl editor which is also open source.
anyone who wants to learn how do do something only have to look at the class in the file format, and or set break points

I can asure you, you will learn one hell of a lot more by steping into the file archive and see hwo a scene is made than by studing the demos,
because the demo remain stagnant in time.
It takes too long for some one to update then or make new ones,
while the file will allow for you to game knowledge from someone else work anyone did using ogrenewt if they want to share thier work.

[Julio Jerez]

F3 should work

Thank you

[Slick]

The code is standarized form app to app, no the mess we are gutthiong now when avery open use hsi own method and some tine we teh get it right and soem time teh getti wrong.
Plaus the file can be seem in the levl editor which is also open source.
anyone who wants to learn how do do something only have to look at the class in the file format, and or set break points

ok good I was confused - so I will be able to step through code to help understand how it works. I had a vision of just loading an example to look at it but not understand how it was done.

[Leadwerks]

Newton scales extremely well. Every single tree and plant in this video is a colliding object:

[razi]

Hey, nice video , I watched a few more videos and see how powerful newton can get. About such big scene, do you create collision objects for such big scene at once at startup or do you maybe somehow stream them?

Also, Julio, how is the work with ogrenewt going?

[Julio Jerez]

I am trying to get a feacture that I thught I was polished and working, it is not.

I wrote a Delaunay Triangulation base on the Barry Joe incremental algorithm without flipping edge.

I use this for decompsiung simpel solid meshed isn simple trehadra solid, teh I can make authmatc compuond tah are very hugh quality.
nut Thei siwll als be use fo Soft body silumulation.

However last week when I try to make a autmatic compund for teh OgreHead and teh Athena in teh Oger Meadi forder teh both no only failed but teh generated ton of slive and nedded son sensical solid.
I am tring to see what is wrong, but for what I see teh Algorth is flowed and for complex mesh with many close vetex.

Ths probelm is challneg me to teh core of my math skills, but I a poised to have it working
I am working of it now and so the I can release SDK 2.25 with the new OgreNewt.