Newton 2.0 Public Beta

[nullsquared]

http://www.newtondynamics.com/forum/vie ... f=9&t=4922

Beta 19. Lots of cool stuff.

- multithreading
- crazy performance difference from 1.53, even with a single thread
- ray cast, convex cast, and multi-body vehicles
- extremely powerful custom joints (you can do pretty much anything with these, not just your classic constraints)
- hundreds of bug fixes from 1.53
- improved contact processing (single callback takes care of all contacts)
- separation of collision utilities from dynamics, Newton can be used as a separate collision engine with no dynamics
- boxes, spheres with non-uniform scaling, cones, capsules, cylinders, chamfer cylinders, convex hulls (any # of vertices), 'concave hulls' (any # of separate convex pieces), non-dynamic (but not necessarily static) tree collision ('trimesh' as others call it), non-dynamic custom user collision, non-dynamic terrain collision
- 'kinematic' bodies - currently requires some slight manual work, but any non-dynamic body with 0 mass (including tree, terrain, etc.) may have a velocity and accurately push stuff around (useful for elevators, trains, etc.)
- dynamic non-uniform scaling of any dynamic collisions (including the 'concave hulls')
- 32 and 64 bit Windows, Linux, and Mac support, as well as Ipod and Iphone support
- lots of other stuff

Interesting post from Julio Jerez:

So of the things you listed the only real issue feature is the generic concave mesh.

The performance, I believe you are testing version 1.53 with have a linear to quadratic solver.
I am going to refer you to at least test 2.0,
The solver ha linear time complexity with mean is about one order of magnetos faster for large scenes.
It is no better in all count that the exact solve because the inner loops have lot more steps, than the general solver.
The general solver had also being made faster in 2.0 and more stable.

For the general concave mesh, I am still working on that. The problem with general concave mesh is not just how to find the face a shape is colliding with, the biggest problem is to determine the quality of the contact generation.

The reason this is and unsolvable problem is because it is a unfishable mathematically,
Mathematically the definition of a contact in the point and the surface normal and the point at the instant tow body touch each other.
The thing is that for a concave edge they will never happen.
This sound like and easy Lemma, but it is much more complex than you think. This is because even those you know that concave edge do not generate contact the fact that the physics is evaluated at discreet step will make an intersection algorithm to find edge contacts. When it does happen you need to provide a solution. No heuristic will find a solution because this is a non exiting problem therefore with does not know how an body should react when it hit a concave edge because it is and even that never happens.

Given that fact how do you solve the issues, the solution is to do back tracking and find the real point at which the body intersect for the first time and collide the tow face normal’s at that point.

Her is the difficulty when you gave two generic shapes this became a parametric N body problem, an not CPU can deal with that efficiently.
The only way to deal with that I know is by using brute force with a large number of processors.

I have seen how in the world of eth self appointed experts the run around claiming the functionality, just because they can do a geometric integration on two polygonal soups.
Then you can see how these features gradually fade away and are discourage to the user.

These people are a group of intellectually dishonest individual’s the use this kind of deception to attract casual user with the purpose to spread propaganda.

This is how you see these competing open source collision physics and collision with a large set of features that do not really work, but that make a lot of noise amount the casual users.
I will never do that.

However I am looking at the new GPU like ATI and Nvidia CUDA, and believe me all those features that require brute force CPU power like cloth, soft bodies, destruction, generic concave collision will be implemented. Make no mistake about that.

(I used bold to emphasize some intriguing points)

[Kencho]

w00t! This really looks tasty

[jacmoe]

Interesting post from Julio Jerez:

I have seen how in the world of eth self appointed experts the run around claiming the functionality, just because they can do a geometric integration on two polygonal soups.
Then you can see how these features gradually fade away and are discourage to the user.

These people are a group of intellectually dishonest individual’s the use this kind of deception to attract casual user with the purpose to spread propaganda.

This is how you see these competing open source collision physics and collision with a large set of features that do not really work, but that make a lot of noise amount the casual users.
I will never do that.

(I used bold to emphasize some intriguing points)

Newton would be a lot cooler, if Julio Jerez weren't such an ass.

Still, you can't really blame innocent bits and bytes, so that's good news for Newton Dynamics.

[nullsquared]

Newton would be a lot cooler, if Julio Jerez weren't such an ass.

I don't think he's an ass. What he said about dynamic trimesh collisions is 100% true. I don't see any physics engine recommending use of dynamic trimeshes, if they even support it.

In fact, he's barely an ass. He responds to practically everything on the forum, and he fixes reported bugs pretty much immediately.

[KungFooMasta]

From what I've read while browsing the forums, there doesn't seem to be an officially supported Character Controller? Writing and perfecting a character controller seems like a pain.

[nullsquared]

From what I've read while browsing the forums, there doesn't seem to be an officially supported Character Controller? Writing and perfecting a character controller seems like a pain.

Character controllers depend on the game. I, for one, am glad there's no default character controller, since my character can seamlessly wall-walk, etc. The usual ray-cast cylinder isn't go to do that well at all.

Edit:
Actually, there is a character controller, I don't know what state it's in, though.

[jacmoe]

In fact, he's barely an ass. He responds to practically everything on the forum, and he fixes reported bugs pretty much immediately.

Well, he is. If it's not about Newton and how glorious it is, he's really gross.
Read this, and burst your bubble:
http://www.bulletphysics.com/Bullet/php ... &f=&t=1334

[nullsquared]

Because August 2007 was yesterday, right?

[jacmoe]

Because August 2007 was yesterday, right?

Is that really all you can say? Did you *read* the topic? I mean, he actually managed to get himself banned. And it's not the first time. Not very likely that he would have changed since August 2007; the quote you posted clearly shows me that he hasn't changed one bit.
Don't get me wrong, Newton is a great library.
And Newton 2 is highly anticipated.

[nullsquared]

Because August 2007 was yesterday, right?

Is that really all you can say? Did you *read* the topic? I mean, he actually managed to get himself banned.

Yeah, I just finished reading it.

I'll have to side with him. Yes, there was indeed some 'childish' behaviour, but the others acted as your usual elitists. Whenever Julio posted something about how physics work, they ignored him and told him how he's being immature, childish, etc. "I believe you have some good physics knowledge, if only you weren't so immature" is pretty much all they said in that thread. And then instead of proving anything, they banned him.

Not sure what part I'm supposed to be noticing Julio's "evil side." What I quoted in the OP seems to be very much true, so I'm not sure what bubble there is to burst.

[Kojack]

These people are a group of intellectually dishonest individual’s the use this kind of deception to attract casual user with the purpose to spread propaganda.

Concave collision is a conspiracy by the physics community to spread propaganda?

Every time I read one of his "the bullet physics lib killed my parents and ate my cat" rants inserted into something like a progress report, if makes me wonder why I still support a newton wrapper in my framework. No matter what the quality of something is, things like this can drive people away.

Saying something like:
"Many physics libraries have attempted concave collisions in the past, but all have had issues with stability and accuracy."
would be a better way to word it. He should hire a PR person to handle these announcements.

As a responsible admin, I should just lock this right now before it gets out of control. Or just sit back and wait for Xavier's response.


Anyway, good work on the progress. Sounds like 2.0 is coming along well. Hopefully soon it will actually be released to the public.

[jacmoe]

I have seen how in the world of eth self appointed experts the run around claiming the functionality, just because they can do a geometric integration on two polygonal soups.
Then you can see how these features gradually fade away and are discourage to the user.

Those self appointed experts would be Pierre Terdiman, Erwin Coumanns, and just about every other expert in the field of physics/CD.

These people are a group of intellectually dishonest individual’s the use this kind of deception to attract casual user with the purpose to spread propaganda.

Also known as the competition.

This is how you see these competing open source collision physics and collision with a large set of features that do not really work, but that make a lot of noise amount the casual users.

Good thing that we have Newton, then.

I will never do that.

Yeah, right.

[jacmoe]

He should hire a PR person to handle these announcements.

Yeah, you're right.

[nullsquared]

Saying something like:
"Many physics libraries have attempted concave collisions in the past, but all have had issues with stability and accuracy."
would be a better way to word it. He should hire a PR person to handle these announcements.

Then you can see how these features gradually fade away and are discourage to the user.

Hey, sounds like the truth to me. I guess people just don't like how he words things.

Sounds like 2.0 is coming along well. Hopefully soon it will actually be released to the public.

Uh, hello, it already is, that's the whole point of this thread. Check the OP.

He should hire a PR person to handle these announcements.

Yeah, you're right.

Well, yeah, he does have some issues with English when he posts stuff. That can definitely be considered a disadvantage.

But, don't lock the thread. It got offtopic. The topic is Newton 2.0.

[jacmoe]

I guess people just don't like how he words things.

You nailed it. Completely. And don't make the mistake of blaiming it on his English.

Anyway, yes: It's cool. But it still is Beta, right?
It's been beta for quite some time now, hasn't it?

[nullsquared]

I guess people just don't like how he words things.

You nailed it. Completely.

But how does that make what he says any less true?

[jacmoe]

Well, insults are not exactly inviting? Especially in a forum frequented by all the top players the physics world.

I picture Julio as a mad professor, who really shouldn't be allowed outside of his lab.

He has created a wonderful physics library, though.

So, yes: great news it is.

[nullsquared]

I picture Julio as a mad professor, who really shouldn't be allowed outside of his lab.

LOL that actually sounds slightly accurate I must say, regardless of the fact that I'm on his side.

[Kojack]

Uh, hello, it already is, that's the whole point of this thread. Check the OP.

Ok, so the forum has the download link. My mistake.

I went to the logical place for downloading a physics lib: the download page. It's still not updated since 2006. I assumed things were still operating as typical on there, everybody needing to ask julio for personal permission to try the beta. This is beta 19. Were any of the previous 18 betas available as publically posted links on the forum, giving us a reason to expect a download if we go to the forum for this particular release?


See what happens when you post his anti-bullet rantings as part of a beta announcement? It distracts those of us who haven't drunk the flavor aid.

[nullsquared]

Uh, hello, it already is, that's the whole point of this thread. Check the OP.

Ok, so the forum has the download link. My mistake.

I went to the logical place for downloading a physics lib: the download page. It's still not updated since 2006.

Yeah, the site hasn't been updated since forever. That's why I have a link in the OP

I assumed things were still operating as typical on there, everybody needing to ask julio for personal permission to try the beta.

It's not exactly personal permission, he simply added you to a mailing list.

This is beta 19. Were any of the previous 18 betas available as publically posted links on the forum, giving us a reason to expect a download if we go to the forum for this particular release?

No. That's why this is stickied as an announcement. That's also why I linked to it in my OP

See what happens when you post his anti-bullet rantings as part of a beta announcement? It distracts those of us who haven't drunk the flavor aid.

It's not exactly anti-bullet. It's 100% true that Bullet and other physics libraries advertise certain features (like trimesh-trimesh dynamics), yet don't advise the user to actually use them, or tell the user to use them "with caution" or "within limits" or something of the sort. Anyways, enough of this; people don't like the way he words things, yet what he says is not all bullcrap, so let's get on with discussion Newton 2.0 itself.

You guys tried the demos in the SDK? Also, the car demos he linked to are pretty neat. Need some more wheel traction, I'd say, and with a couple of more tweeks it'll feel Just Right.

[xavier]

Because August 2007 was yesterday, right?

In Julio's case, the tiger cannot change his stripes. Julio has done nothing to demonstrate that he is a different person than he was in 2007.

[xavier]

No matter what the quality of something is, things like this can drive people away.

And do.

[xavier]

It's not exactly anti-bullet. It's 100% true that Bullet and other physics libraries advertise certain features (like trimesh-trimesh dynamics), yet don't advise the user to actually use them, or tell the user to use them "with caution" or "within limits" or something of the sort. Anyways, enough of this; people don't like the way he words things, yet what he says is not all bullcrap

The issue is that Julio continues to destroy his own credibility when he says these things.

Havok, for example, makes no such warnings -- for example, use the MOPP all you want (and we do), it's perfectly fine, perfectly stable, and I've yet to have the problems with capsule/trimesh collisions with Havok that I had with Newton.

I don't need Newton -- many don't. Havok is just as free (on PC, which is all I care about at present) and developed and maintained by many of the best engineers in the field. No matter how bright Julio may be, he is still only one person, and can do only so much.

And Newton still has no tools.

[Kojack]

No docs? At least the demos are commented, but a doc would be nice.

The demo needs a little fixing. It's framerate display always prints 59.999996, even when fraps measures it at 16fps (verified by the demo's own profiler graph).
When you grab a skinned ragdoll with the mouse, it's either having an epileptic fit or dancing to techno, it jitters around. Pick it up by the torso or arms and you can lift it easily. Pick it up by the head and it's too heavy to get off the ground. I'm going to have to assume that the mouse grabbing doesn't apply a consistent force, because currently they don't behave realistically. Although maybe that's why every ragdoll related function in the main header is commented out. In fact, all vehicle functions are commented out too.

[_tommo_]

But Havok is free only for non-commercial use?
Anyway it's sure a step before the others, i remember in 2005 creating 64 Dr. Breens and making them explode
Also if that was an hell of ragdoll-ragdoll-geometry collisions it never had an issue, on an old P4... I don't think that today there's another engine capable to do that.

In fact I tried newton only once, and it looked much less "professional" than ODE at the time, with only spheres and boxes... anyway it looks much better now, but it's nothing epic i think.
PhysX is still good for me, even if Nvidia hasn't updated it for ages... it was really advanced before, but now the latest feature was implemented in 2007... I mean, it still lacks Cylinders.