Help Surface-based collision option not working

[budosuyu]

It doesn't seem to work for us. Ticked or not it works the same way it did with the previous versions. Am I doing something wrong? Are there any additional settings I should be aware of? 

I wish there were a demo scene specifically for this new feature as we have seen in the new demo videos.

Can you please help me out?











GIFS: 

[josemendez]

Hi there,

If you created this rope in previous versions, make sure you trigger a re-generation of the rope blueprint as explained in the manual:

http://obi.virtualmethodstudio.com/tutor...sions.html

Existing pre-6.0 actor blueprints will need to be re-generated for surface collisions to work. In case of ropes, changing any rope parameter in the path editor will trigger a blueprint re-generation. Keep this in mind when upgrading existing 4.X or 5.X projects to 6.X.

[budosuyu]

Hi there,

If you created this rope in previous versions, make sure you trigger a re-generation of the rope blueprint as explained in the manual:

http://obi.virtualmethodstudio.com/tutor...sions.html

I only exported nothing but the models and materials and imported them to a fresh project with obi 6. It doesn't contain any trace of the previous obi. Created ropes and blueprints from scratch.

[josemendez]

I only exported nothing but the models and materials and imported them to a fresh project with obi 6. It doesn't contain any trace of the previous obi. Created ropes and blueprints from scratch.

In that case, surface collisions should work. I can't reproduce any issues.

In your screenshots it's very difficult to tell whether they're working or not, since particles are overlapping each other so there aren't any gaps in between them. At lower resolution, the ropes should collide with each other even if there's gaps. Generally, it doesn't make any sense to use surface collisions for high resolution ropes, as explained here: http://obi.virtualmethodstudio.com/tutor...sions.html



Note that surface collisions won't prevent tunneling in any way: they perform spatial supersampling, not temporal supersampling. Tunneling is prevented by using a smaller timestep.

[budosuyu]

Does this give you an idea if it is working as intended or not?

[josemendez]

Yes, it's working as it should.

However the ropes seem extremely overstretched (I can tell by the particle pairing, that appears when using sequential evaluation for distance constraints), so constraint resolution will force the ropes to pass trough each other in order to meet distance constraints.

The solver strives to meet all constraints simultaneously. If you stretch the rope that much past its rest length (so that distance constraints are already being violated), collision constraints will only be enforced to an extent, as fully enforcing them would further violate distance constraints. No temporal or spatial supersampling of any kind will prevent this. In the real world, the rope would have snapped way before reaching this point.