Destruction 2.0

I’ve seen that the destructible related points (like Destruction improvements, Destruction Optimizations, Dynamic fracturing were removed from the Roadmap because “destruction 2.0 replaces this effort”.

The description of Destruction 2.0 is:

"A new destruction system that moves beyond pre-fractured rigid bodies, to allow real time fracture and deformation. Possible based on particle/FEM physics, maybe GPU accelerated."

Could you give us any more info about how this will work and when we can expect to see it being implemented?

I’m also interested, sounds neat!

I would also like to hear more about this.

Nice to see that I’m not the only one interested in this :slight_smile:

As you can see from the ‘backlog’ label, we don’t plan to work on this any time soon I’m afraid! The description was probably a little ambitious, I shall scale it back on the next roadmap update. The initial focus will be on reworking the existing DestructibleMesh system, in conjunction with NVIDIA, to have less overhead and be more efficient. It would be nice to support dynamic fracturing eventually, but we want to improve the performance of the existing system first, and need to do some research to see how feasible that is (e.g. vertex memory usage). After speaking to some teams who do offline destruction for vfx, we probably will stick with rigid body (ie not particle/FEM).

Thanks for the info!

Working on the current system would already be a huge improvement. Especially if something like performance of spawning DMs with “Form Extended Structures” enabled is improved, that can cause huge frame spikes at the moment (UE-16792). I have reported quite a lot of bugs regarding the destruction system during the last year, and I’m eagerly waiting for a time when they get addressed :cool:

Shipping a game that makes heavy use of destructibles is quite hard at the moment.

@Epic: Pretty nice info! Can’t wait to see it coming.

Here is a small snippet i wrote to Hobson a couple of hours ago, maybe you can help us out on this:

A) - I read a lot about the problem that there is no way to dissolve/fade out the chunks.

Is it possible to expose the chunks inside a BP and then adding a timeline with opacity function to it?

Alternative way 1: Override chunk’s material once it’s active to use the whole chunk (inner + outer mat id), we may wanna use a dissolve shader on it then!

Alternative way 2: Shrink chunks over time. It worked pretty well inside Cinema 4D (native 3d application, using an explosion modifier)

B) (If there is no solution for “A”.) Destroy chunks on second hit from a weapon? (first, i used “crumble smallest chunks” but this is not giving the desired effect of “fading/destroying” chunks)

C) Separate Chunk collision from initial DM. Our ships are either collide with both or no collide at all. (Large chunk threshold not working on our physic-ships)

At the moment i circumvent most issues above by killing the chunks in under a second replace them with particles.

So does that means that a technique like this one will be used instead ? , i think that is even already in the last version of apex .

http://physxinfo.com/news/11123/physx-research-real-time-dynamic-fracture-explained/

The trello card has been archived, what does that mean? Trello

Probably, NVIDIA Blast is going to be the “Destruction 2.0” and replace APEX Destruction

Especially since Blast is promised to be in “UE4 Mainline” alongside with nvCloth, which is already integrated in 4.16

Where you see the refferance to blast, I can’t see it on the timeline.

The latest I read from Staff is that, currently they have no plans for Blast. However Destructibles definitely need an overhaul, being it Blast, Bang or whatever. The number of bugs and limitations of the current implementation is… outstanding.