PhysX Discussion (dev branch)

[dragon01 902]

Since this has been bumped anyway (been meaning to do it myself), I'd like to ask what happened to the fix for amphibious tracked vehicles. It's been promised for Apex and while it's understandable that it might have been delayed a bit due to expansion's own problems, I'd really like to know what are the current plans regarding this. Several mods would be really grateful for that fix.

[x3kj 1247]

It's been removed for "regular" devbranch and stable. For whatever reason... only using devbranch arma3diag.exe the feature still works.

On another note - BI, you should remove the fake gears that where added to tanks long ago. I found out (while working on my own vehicles) that adjusting the torque curve for the engine is enough to get the vehicle moving properly, even on steep inclines. Removing fake gears reduces the awefull sound due to rapid changes and also leads to less gearhopping/stuttering when turning.

This works very well for me:
torqueCurve[] = {
{0.000, 0.000},
{0.265 ,0.600},
{0.417 ,0.800},
{0.663 ,0.982},
{0.758 ,1.000},
{0.850 ,0.930},
{0.925 ,0.807},
{1.000 ,0.600}
};

 

Idle rpm is at 0.23 in my case. Torque at idlerpm is important.
With this i use gear efficiency of 80-85% in lower gears and 90-95% in higher gears -> less gears required and also less gears for the "autobox" to choose from ->smoother ride experience.

[Redphoenix 1540]

But this is the only curve that works?

[.kju 3244]

Will tankX simulation/physX ever get improved from the sad state of A3?

Or can you please fix the tank/non physX simulation again finally instead.

[x3kj 1247]

But this is the only curve that works?

Obviously not. But it's important that there is torque at idle rpm, otherwise you won't get the engine to spin up -> no torque -> fake gears required. Also, you notice that my curve is a bit more flat near the 1.0 torque area , but still with enough falloff at beginning and end. A flat section is important so that the efficiently usable range is not only a a few hundred rpm (-> less gears required, less switching required), and the falloff at beginning and at end of usable rpm range is important so that the autobox is definitely forced to switch.

idle rpm in my case is 0.23 of max rpm

 

Will tankX simulation/physX ever get improved from the sad state of A3?

And if not for A3, please put some work/research into it for enfusion at least :/

BI, if you need assistance with the technical background/theory, i'd gladly help. I'm an engineer and did quite a bit of research into tracked vehicle movement and groundinteraction by now.

Another note on little "breakthroughs" i made - I noticed that the problem of vehicles completely stopping when they drive into a small dip is gone when one makes the physx wheels larger than they would be in RL (moving the wheel centerpoint upwards). Exterior visuals are unaffected by this, and behaviour is basically not differentiable between the two.

[kklownboy 43]

Ok not sure if this is the best place for this info;   PhysX 3.4 on github.     https://developer.nvidia.com/physx-source-github

 

And a good blog on it;

 

http://www.codercorner.com/blog/?p=1257

 

the January archive;

 

http://www.codercorner.com/blog/?m=201701

 

 

[x3kj 1247]

from tank dlc thread

On 21.10.2016 at 8:21 PM, x3kj said:

• Solve the sudden jolts that propell the tank in various directions. Implement a limiter to acceleration/forces/torques inside the physics solver. A tank accelerating from 0 to -50 within 1 frame is not a thing that can reallistically happen.

From http://physxinfo.com/files_new/PhysX_3-4_release_notes.html

Quote

A new function PxVehicleSetMaxHitActorAcceleration has been introduced. This function sets the maximum acceleration experienced by a PxRigidDynamic that finds itself under the wheel of a vehicle.

Almost.... almost...