Scaling active objects results in inconsistent collision detection

[Dominic]

Is this just the way things are? Collision events will just randomly not go through when dealing with objects that are scaled. Is use of precise collision known to make things worse in that regard? Help me figure this out!

[ProdigyX]

What movements are you using for the objects? Objects with physical movements do not get their hitboxes updated when said object is scaled. As such, physic-based hitboxes are currently unaffected by object scaling.

[schrodinger]

Bumping up this old thread because I've just encountered this problem I was not aware of.

I'm dealing with it in a custom-movement scenario,
I'm checking overlaps between objects on loop, to negatively-position them when overlap occurs.

I just noticed that this same collision code works when object's scale is 1,
and then completely stops working (object passes through) when scale is different from 1.

Any suggestions on how to deal with it,
or any possible fix for this issue in the horizon (...how soon...? )

I really need colliding with scaled objects.

Going to make further testing meanwhile.

[schrodinger]

Mmmm, I'm failing to reproduce the issue in a simpler environment.

I'll keep trying to isolate the problem or identify some other cause in my main code...


UPDATE:
same exact code works ok in a brand new example and not works in main application.

Only difference in main application from working to non working objects is scale <>1.

Going to be a long night...


UPDATE 2:

Setting back scale to 1 while running application makes not working object's collision work again.
Setting scale of 1-scaled objects to 0.9 or to 1.1 breaks collision.
Gosh.
This can't be a coincidence..
(I know, you may very well not be interested in this )

UPDATE 3:

this is driving me crazy.
<>1 scaled objects colliding ok when "use fine detection" is unchecked.