UE4 has been out for just over a year (yeah, in development for longer but it’s still young).
You could (not-easily) change the Physics Engine. All of the callbacks to PhysX are wrapped in #define tags, so you could create your own functions wrapped in different #define tags and call to another physics engine. Good luck with that though…
Only the starting positions, not the updated positions during simulation (e.g. ragdolling joints). PhysX does not offer this.
Regarding Softbody uses:
as above: BeamNG
and: Spintireshttps://youtube.com/watch?v=qiOt7SuHGPM
But there is the Tendr PlugIn coming. Hopefully this will save the day. Any driving game benefits from soft body physics.
My main quirk with UE4 is the non existent physics multi threating. This is not good.
I probably came across as a bit dismissive, but the use of soft body is kind of a very limited use feature with the majority of games. So I think its unfair to call out UE4 for not implementing them. Car games have been doing destruction of cars for a while without them, so yes its a nice addition, but likely so project dependant that it would need Epic to actually want to make racing games for them to need to implement it.
Again to the OP, if there’s feature you want, go ahead and implement it. Thats why you get the source.
Hell, you even get the source to Physx these days too. And PhysX IS multithreaded last time I looked, so are you actually testing this? If so, maybe try and determine WHY its not working in multithreaded mode, rather than complaining about it?
i spoke to beamng devs some time ago and they said an sdk will be available at some point but have no set date.
yes physx (specially the current version) was a bad choice for ue4 physics but its too late now to ask epic for something else.
hopefully someone will rip it out completely and replace it with something at least half way decent, and in an ideal world offer their branch to us.
Speaking on PhysX Multi-threading you can find a post on this thread with our Physics Engineer: Please use PhysX 3 :) - Feedback & Requests - Epic Developer Community Forums)
specifically starting with this post: Please use PhysX 3 :) - Feedback & Requests - Epic Developer Community Forums)&p=225359&viewfull=1#post225359
This is certainly a valid argument. Dwelling into the PhysX Code however is far from a 5 minutes job, as already mentioned above.
One could draw the comparison of buying a car - but all disassembled - and the car dealer says ‘what do you want, it is all there. You got all the source parts’. 
The entire point of using UE4 is to not do things like that yourself, in my opinion. I could write my own little engine otherwise.
Physics multithreading is essentially included in any engine by default, because it is a very cpu stressing application.
Yes, PhysX itself is multi-threaded, it just seems the UE4 task scheduler does not distribute very well.
Is Ori Cohen referring to the async scene? When I run physics in UE4 on my Intel i5/Win8 and the simulation gets heavy, then the frame rate goes down to slide show level and there is no sign of using more cores. So usually 2-3 Cores of my quadcore are not used. (no async included, which is destructibles limited I believe) and I look at a slide show.
Really, this is the reason why Ubisoft and other publishing studios / developers choose to build their own engine from the ground up? Not the fact that, in the long run, it’s more financially viable? Ubisoft’s used the their Anvil engine (sure, they’ve iterated on it consistently) on all of their Assassin’s Creed titles.
Whereas, if they had utilized Unreal Engine, they’d have to pay licensing / royalty fees and so forth on all those titles. They’ve cut out that process, by using their own. Even if it costs them more in the short-term, they can re-use the same engine on numerous other projects without paying Epic any cash.
In the end, there’s numerous reasons why developers choose to develop their own engine or licence an existing one. So, it’s best not to make sure blanket-face statements. As, what works for one, won’t work for everyone.
I can’t quite wrap my head around this statement right here. There’s a whole host of free engines out there, along with others which offer a relatively long-cost entrance fee (even CryEngine, for example). They aren’t associated with ‘amateur zombie shooters’ to my knowledge. In fact, I think the whole zombie craze has been slowing down a fair amount as of late - at least, in comparison to a couple of years back.
However, have you seen any of the work that’s coming out of indie studios that are harnessing Unreal Engine 4? It’s not quite ‘amateur zombie shooters’, to say the least. In fact, far from it. You only need look at some of the recent titles that have been given grants from Epic or take a peek at the work in progress sub-forum.
Is there going to be zombie titles using the engine? Sure. Are some of them going to be amateur? Sure. That wouldn’t change if the engine was free or maintained its former subscription fee. Is that what the engine will be known for in ‘one or two years’, as you say? Certainly not.
To sum-up, though. You should find what works for you. If UE4 doesn’t fit your criteria and CryEngine does? Then move to CryEngine. It seems relatively simple. If not, work on implementing what you feel is missing in Unreal Engine.
Softbody tires/mud sim in Spintires are custom code, afaik, and not Havok.
Except for custom tires, you can build complex constrained vehicles (http://www.youtube.com/watch?v=D8zQQ2Rm0ow) and deformable terrain (http://www.youtube.com/watch?v=bLIBE9bsfn8) with PhysX.
a) BeamNG soft-body tech was in development for many years (including Rigs of Rods lifespan), and it is very specific and vehicle oriented. I doubt a general soft body plug-in will give you what you want.
b) BeamNG-style destruction does not fit many games. BurnOut style game with instant respawn or ultra-hardcore racing game - yes, other - no.
And this “something” will be ?
if i had the skill and time to replace physx it would probably be with bullet, or if/when beamng release an skd it would be that.
The entire point of UE4 being released with full source, is so that we can implement features that we may require for our game without having to wait for Epic or going without. Sure I could write my own engine, but why would I, if I only require, for example, soft bodies, I’d rather just implement that into UE4 myself. If you don’t have the skill to do so and still require a specific feature, you could post a request in the forums and get someone who knows how to code to implement it. Or wait.
Significantly slower than PhysX, no vehicles, no clothing, no destruction, no detailed documentation, no DCC plug-ins, no debugging tools, very slow development rate, no proper support and optimizations for consoles ?
It is easier to modify PhysX source.
?! Very specific physics engine, that can be used only in racing games. It can be utilized alongside PhysX, but not replace it.
beamng has been developed for vehicles but you are not seeing the bigger picture if you think thats all its good for. thats the same sentiment as ue4 is only good for 90’s zombie shooters.
bullet has vehicles and cloth and destruction and softbody and documentation and is cross platform and free and open source. seems an ideal replacement to me…
no point in arguing over what if’s though, ue4 is stuck with physx at least for the foreseeable future.
as for implementing features like softbody ourselves, if the people who made physx cant even get it working properly what chance does anyone else have.
NVIDIA Flex is already shaping up to be pretty good at it, then theres the Tendr plugin also shaping up to be pretty good, so theres two solutions already, and god knows how many that haven’t been publicly announced.
Nah man… i just like the chirp once a while 
If it seems like i was directing that at you… i wasn’t. I was simply commenting on soft bodies.
But i agree. No need to put the engine down. Cryengine has features i would love in UE4 like propper river tools, water volumes, fog volumes etc. But UE4 has a **** load more features i would like to see in cryengine. Like blueprints…
And blueprints technically count as 50 features imo because i cannot code and its essential to my development cycle.
*Note BeamNG use the Torque3D + custom physics, no the Cryengine.
Deformable Objects + Boolean Destructibles will be a nice add to the UE4, plus if Epic Games give the option to select between 2 physics engines (PhysX & Bullet) that will be . (Since Bullet is a open-source lib, and UE4 have a lot of Open-source libs, but have 2 physic engines need much rework and additions to the code).
About “Nothing good in life is for free” then UDK is bad ?
*A huge amount of libs used in this and others engines (private or public) are open-source libs, then all engines are a trash ??
There are other games with that physics example of “Next Car Game” (other cars example)
But the soft physics can be used to make deformable barrels or things static or dynamic or with physics over the map.
About the Tendr Dynamics plugin Tendr: a real-time, volumetric, soft-body physics plug-in for UE4 - Game Development - Epic Developer Community Forums this plugin have a AGPL license, and i no idea about if that is usable with UE4 for public projects. (I think not)
Just to be fair:
Not until you code them on top of Bullet raycasts. PhysX has complete Vehicle SDK.
Which is inferior (in both performance and stability) to specialised APEX Clothing solver. Plus no DCC tools. Say goodbye to character clothing - most commonly used cloth in games.
Nothing that can be compared to APEX Destruction. No editor, no LOD system, no rendering optimizations. In Bullet, you’ll need to write your own system from scratch.
An old position based algorithm, like in PhysX 2.8. FLEX is much more promising.
But does it have the amount of platform specific optimizations like PhysX ? I doubt it, simple because it has much smaller development team behind it.
Too bad such ideal engine is used in actual games so rarely (name me at least 20 of them)… maybe something not right here ?
And, yet again - Bullet is slow (some benchmarks below)
http://www.garagegames.com/community/forums/viewthread/136369
http://www.codercorner.com/blog/?p=761
a) Bullet 3 will have, because it is not released yet
b) What it will really have - a tons of bugs. A complete new engine, written by a few people, not tested in any games yet. Not a smart choise for integration.
Great… so instead of one good integration, let’s have two bad ones. Because we have nothing else to do =)
And will Bullet have a level of developer support NVIDIA is offering here ?
As for softbodies, why I see here only hyped stuff like BeamNG ?)
There are more promising and universal engines out there, like Numerion Carbon. Just look at those demos !
http://www.numerion-software.com/carbon-demos
Too bad thing is already half-dead.
And have everyone already forgotten about Digital Molecular Matter ?
There was so many hopes when it was announced)) Now the real-time engine development is suspended, only VFX version is updated.
Why is that guy vandalizing that hallway? Those artists spent days creating those assets and now just because he some jedi scum he goes and break everything 
that is catastrophically broken
yes thats cool but doesnt cover deformation
yes but it doesnt work yet
my guess is nvidia have a better sales person, the games bullet is used in work fine
no the point is to have 1 good one, havok is way better than bullet but the full version costs too much, hence i said bullet
because its freaking
flex is not yet fully implemented and caters more for fluid dynamics, Tendr has not been released
its obvious ue4 is an engine aimed squarely at the future and thats a wonderful thing. unfortunately my time machine’s max speed is only 1 second per second, leaving me in what can be described as the present.
saying this physics is better than that is neither here nor there.
this forum section is here with the intention of providing feedback on ue4 so its developers can make it even better.
this is just one of many many threads voicing justifiable concerns over the current physx implementation. if nobody makes epic (and nvidia) aware of the problems nothing will be done about it.
When you can’t create a good game you have to resolve in bling-bling (soft-bodies in this case).
Gran Turismo managed it quite well 16 years ago…
ue4 is all about the bling man 
or havnt you noticed…
if i compared ue4 to a diamond watch,
yeah it looks cool but doesnt actually work, half the diamonds are missing and ive had to tie it on with a piece of string.
