Modify contact_max_allowed_penetration precision to 3 significant digits

[Malcolmnixon]

This pull request addresses #75634 by modifying the numeric precision of "physics/3d/solver/contact_max_allowed_penetration" to three significant digits, so values down to "0.001" can be specified. This is necessary in VR situations where the scene contains objects at the scale of human hands.

Bugsquad edit:

• Fixes [GodotPhysics] RigidBody3D Objects Overlapping #75634

[capnm]

Is there a reason why a value 0.0 is also valid ?
physics/2d/solver/contact_max_allowed_penetration - has, I guess, the similar issue(s) ??

[clayjohn]

Seems fine to me. A minimum value of 1 mm seems more reasonable to me than a minimum of 1 cm

[fabriceci]

Did you look if it doesn't fix other issues (jitter on small objects)?

[rburing]

Would you mind adding a Co-authored-by line in the commit message?

[rburing]

Looks good. For 2D no change should be necessary as the value is in pixels.

P.S. My e-mail address is public, and can be found e.g. on my website or by using git log on a local Godot repo.

[Calinou]

@rburing Is a value of 0.0 legal? The property hint currently allows it.

[rburing]

@Calinou Setting it to 0.0 means no penetration is allowed, so the engine will try to depenetrate (rigid and soft) bodies pretty much constantly, which is almost certain to cause constant jitter.

So we could set the minimum to 0.001 to be a bit more user-friendly.

Maybe

• physics/3d/solver/contact_recycle_radius and
• physics/3d/solver/contact_max_separation

also need a similar treatment of increasing the precision.

[Calinou]

In this case, remember to also change the minimum value for the 2D contact_max_allowed_penetration 🙂

[Malcolmnixon]

If we're going to include additional changes to ranges and limits in this PR, I would like to get confirmation on the changes:

1. 3D units are physical distances, and so 0.001 (1mm or 3 significant digits) is appropriate:
   a. physics/3d/solver/contact_recycle_radius modified from 0,0.1,0.01,or_greater to 0,0.1,0.001,or_greater
   b. physics/3d/solver/contact_max_separation modified from 0,0.1,0.01,or_greater to 0,0.1,0.001,or_greater
   c. physics/3d/solver/contact_max_allowed_penetration modified from 0,0.1,0.01,or_greater to 0.001,0.1,0.001,or_greater
2. 2D units are pixels, and so 0.01 (2 significant digits) is appropriate:
   a. physics/2d/solver/contact_recycle_radius keeps its range of 0,10,0.01,or_greater
   b. physics/2d/solver/contact_max_separation keeps its range of 0,10,0.01,or_greater
   c. physics/2d/solver/contact_max_allowed_penetration modified from 0,10,0.01,or_greater to 0.01,10,0.01,or_greater

Is everyone fine with the significant digits choices based on 2D vs. 3D, and that only contact_max_allowed_penetration needs to have zero excluded from the range?

[Calinou]

Is everyone fine with the significant digits choices based on 2D vs. 3D, and that only contact_max_allowed_penetration needs to have zero excluded from the range?

Sounds good on my end, but I can't confirm whether contact_recycle_radius and contact_max_separation should be able to use 0.0.

[Malcolmnixon]

The PR has been updated with the appropriate recommendations.

[rburing]

Looks like a fine improvement as is.

[YuriSizov]

Thanks!

[YuriSizov]

Cherry-picked for 4.0.3.