Scaled-up physics slowdown

this is drivin me insane.

scaled up everything in the scene 4 times the regular size.
the physics are different with these objects
but I don't understand why it plays as if timescale was low.

the speed is
speed = rigidb.velocity.magnitude * 3.6f;

it's not the camera
and its not low fps



has anybody encountered this?

editd: okay it seems to be the expected behaviour...
(and the speed reading is correct but little for this scale)
there's no way around that?

 

Well, scaling is not a good thing in physics in Unity, specially in vehicle physics. You may need to keep track of a lot of scaled things, for example:

• Scale the speed meter by 1/4, so the speed value looks realistic to that size.
• Scale the time by 4 (Time.timeScale = 4) so the cars move 4x times faster.

You'll have to deal with similar issues with suspension, engine (it's spinning 4 times slower), and so on. Not to mention the inertias, which are difficult to fine tune correctly and have a direct dependency of the physical dimensions.

I'm yet to see a valid reason to scale physics in Unity, so I'd really recommend you to use 1:1 scales.

 

yeah, it's a tough issue.

scaling up time didn't work out, but smaller scale 1.5-2 can be fit to alright.
the speed didn't look right to me (when dividing by 4)*, looked faster, but maybe the slowing down effect was playing tricks with how it felt

*see the gif. does that vehicle look like going at 25-30, when it shows 100-120?

I have a feeling something is going on with fixed update / update /. physic loop cycle
added higher motortorque, higher mass ...

the reason for me that subjectively the drifting feels better, easier controllable by input than in regular scale

gamedev is a rollercoaster
sometimes I feel nailed it, sometimes whole thing goes wack
now I feel satistied with the result, with a little bit of work on input, brakes can be good
http://www.mediafire.com/file/xm9fglv6w3iqwbq/RGScene02c.ZIP/file

 

I bet you can get the same effect by using 1:1 scale everywhere, then scaling the rigidbody's inertia.

 

thanks for the tip.

you mean inertiaTensor? because I couldn't find a value that wouldn't throw the car up and down with it.

anyway I just realized I was using timescale 2 in the demo above,
so that means other problems elsewhere as well...