Havok release version 4.0 of it's tools

http://www.havok.com/content/view/341/53/

think we'll ever get anythign close in sl?

I find it especially ironic that the tagline at the top of Havok's website says "Stay ahead of the game"

A little late for the party, we already had the 'haha havok 4.0' Linden bashing party a couple months ago.

Probably not, either. Havok is not really a priority for Linden Lab according to Philip, and they were last working on 2 (although they said they'd probably go to 3 now cause the hooks for the engine are all the same as 2). Who knows though, when they get around to actually making it a priority in 2010 and Havok 8 is out, we might get 4.

I've been talking Havok 4 since the day I joined Second Life (since its progress has been very open to the development world). I am making its need a major part of my presentation at the SLCC.

Hopefully by the end of the conference I can have some idea of what was breaking in their transition to Havok 2 and what kind of resources are needed to proceed to Havok 4 Physics (though I'd love to get their animation engine as well, but that seems unlikely given SL's current, no pun intended, model).

New havok upgrades are not important unless it improves the performance of physics calculations on the server.. Nonthing is running clientside.

Second Life is not a game.

Phil said it's not a priority because they were only looking at upgrading to reduce server side crashes due to physics, but that physics engine crashes were a huge minority compared to other things causing crashes, hence they don't care about physics right now. He also said that even if they did upgrade to Havok 2/3 tomorrow we wouldn't notice much change in functionality.

In the long term though, upgrading the physics engine provides the potential for things such as ragdoll physics and other nifty things. Our current physics engine, Havok 1, was old when they first started using it from what I understand. Now it's approaching antiquity. One has to assume that between Havok 1 many years ago and Havok 4 now that there's been some substantial improvements in general physics calculations as well as shiny new capabilities.

Ragdoll is clientside and one of Philip's favorite lines is saying SL looks the same from any system because everything is server side.

...which it would.

As I understand it, Havok N (where N>1) uses far better algorithms which would allow for:

- finer-grained timestepping (or smaller sim load with the current time stepping)
- smaller size limits for objects (or smaller sim load with the current size limits)
- decreased object interpenetration
- better resolution for object tracking (sim border crossing issues, anyone?)
- greater sim stability in the face of complex physical objects (elimination of the 'deep think' problem where sims get stuck in an infinite loop when trying to determine motions of linked physical objects with nontrivial geometry)

All laudable goals, I would think.

I know that is true for Havok 2, but I've never heard anything about the stepping in Havok 3 or 4

...and thus crap.

If there's one thing that would make me switch from SL, it's the promise of better physics to play with.

Presumably if the tech worked well in v2, they would keep it around (in existing or improved form) in the upgrades too.

Thank you, thank you, thank you.

That's not the point.. I have heard nothing of improvements over v2 in that area. It's pretty much assumed that whatever is in v2 is going to be in v3 or improved in some way. I would like to hear of any improvements over v2 tha are in v3 and v4 that would have an impact on Second Life's physics simulation.

Well, given that Second Life is only using Havok 1, and we *know* Havok 2 is a significant upgrade, then whether they upgrade to v2 or some more current version of the Havok physics engine SL should exhibit the improvements in behaviour I outlined.

At any rate, the big improvement Havok made in v3 was the introduction of "continuous physics", which causes better collision detection. This is particularly important for fast-moving objects, in which discrete timestepping can cause objects to tunnel through one another. This would presumably also have a positive effect on the existing sim crossing problems as relative positions of objects would be determined with far greater accuracy.

What effect would open sourcing Second Life have on the importance or lack of importance of something like Havoc 4?

Seems like one might not be able to say "Who cares, LL doesn't care about physics so we on't get it" if LL wasn't the only ones working on the source.

I wish LL would commit to havign abetter physics engine. A better physicis engine would make sl a better platform for game development, and thus a better all around platform

I'm not sure how they would work Havok licence with OS.. Entirely possible that a new physics simulation would have to be written. The way the system works now though, as long as the simulator knows what updates to send there would be no client compatability issues with a physics change.




If you're interested in an improved platform for development in SL, check out the threads in my sig ;0 You can ignore the blog at the end ;0 It's pretty dull ;0

Has anyone found out anything as far as a LL development timeline for Havok > 1 ??

No.

How did your presentation on Havok go?

Yup.

It's not happening.

Ever.

I'm sorry paulie, i guess you never saw this thread:

/108/1a/129305/1.html

While I proudly remained a sponsor of the Second Life Community Convention, I didn't feel the extra expenses associated with attendance as I had planned it (particularly my publication) were worth it. Today we see that Linden Lab is committed to bug fixes, new and long-standing, as well as their conversion to the XUI client-side user interfce library.

Unfortunately it appears that the back-end is of low priority, since they only appear to be chasing the "social" MMO and "commerce" side of things where there is little competition. I will continue to make proposals and honor my committments to others on my projects.

Given the presentations at the SLCC that I viewed through the video feed there appear to be a lot more projects dedicated to APIs, particularly web-based, rather than upgrades to the core technology. If there appears to be any change in this direction while I'm still around I will definately pursue it and be happy to report back anything I find.

Social MMOery, they *do* have competition, though in terms of user commerce they don't really have any. But in any case, I have certainly come to appreciate personally that the physical world, which is what really drew me into SL in the first place, is not going to be improved in the near future, and that saddens me.

It's not just me either; I talk to a lot of people, old and new, who want to play with physics and interactive objects, and they are stifled by the flaws with the existing system, ones which have existed for years and have still not been fixed. What is the point, for instance, of a physics engine which results in me advising people that if they want even slightly reliable moving objects, they should use non-physical systems? "SL physics is good for things that go boom, and that's about it" is something that I've said many times.

I don't think I can be described as an incessant complainer but this issue really does get my goat, and, as I mentioned earlier on back when this thread was new, if there was one thing that would make me switch from SL it would be a buildable, scriptable environment with a better physics engine.

For Andrew Linden: how's the Havok integration coming along? Any ETA on when we will get "joint" functionality back? Remember when you removed joints because their removal would ease the integration of the new physics engine?