SketchyPhysics 3.5.6 (26 January 2015)
-
@faust07 said:
A typical example...
Thats very strange. When I added option readback collision geometry to the vehicle, there were 3 collision shapes below it, which came out from nowhere. That's some strange effect with default shape. Meantime for the fix, set vehicle shape to convexhull2 or compound2 for proper collision generation. Convexhull2 is preferred as it's the fastest.
I can't reproduce this in SP3.2, so that's some bug by me, which must be eliminated. I will look into fixing it.
Thanks for report.
-
If You add option readback collision geometry to the wheels You will sea that it happens only with flipped objects/groups ...
-
Okay, it is fixed now! Please download latest version and see how it goes. Thanks for report, faust07
-
It works. Thanks again Anton. The door to the "Physics World" in SketchUp is wide open ...
-
-
I think I see what you did there
-
Oh no, Anton I don't want to nerve you. I cheered too soon. A model that ran in SPh3.5.3, jolts in SPh3.5.4 a little ... Probably the steering blocks are squeezed out laterally from the body of the truck. I hope the GIF example shows what a little.
-
@biani patrick said:
Si ça peut être utile !?
Thanks for your documentation, patrick. I placed this in first post, as part of reference.
-
@faust07 said:
Oh no, Anton I don't want to nerve you. I cheered too soon. A model that ran in SPh3.5.3, jolts in SPh3.5.4 a little ... Probably the steering blocks are squeezed out laterally from the body of the truck. I hope the GIF example shows what a little.
Is the behavior different in SP3.2? Are you assigning same shapes to SP3.2 and SP3.5.4? Meantime, if they are squeezed out, to fix it, set most surrounding parts and the cabin to Not Collidable.
-
@faust07 said:
No, in SP3.2 it did run OK. There were only the other reasons for crashes, you've solved so wonderful! In your SP3.5.3 this truck moved correctly (after resetting some mirrored objects).
Can you send me the skp model (private message if you prefer), so I could investigate.
-
No, in SP3.2 it did run OK. There were only the other reasons for crashes, you've solved so wonderful! In your SP3.5.3 this truck moved correctly (after resetting some mirrored objects in SPh3.5.3).
-
Where are the physics joints and the wheels? A lot of parts of the model are deleted. Do you mind reuploading? Don't worry about size. You can erase the file from attachments after I download it.
-
Here it is. It's a model on the base (not to forget to mention) of posts of Mr.K. and other authors for cars following a path (with little modifications I did, to use it universally). I've blown it up with physical walls, doors, roof, seats, driver, wheels, lights for twilight renderer, euro norm ... Hope, you find a solution.
Oh sorry, have caught the wrong file. It was too dark last night and there was too much port wine ... I must go to my other computer to upload the correct file. -
I see what it is. It has nothing to do with the flipped bodies. The problem is that in SP3.5.3 and below the Default shape did not include all geometry as part of collision generation; however, in SP3.5.4 I made it to include all geometry, which I see breaks some compatibility with the prior models. The wheels are pushed out by the lower black part of the cabin which according to Default shape has a box collision. To fix this, you enter the cabin group one level deep, and set the lower black part to ignore:
In prior versions, the lower black cabin wasn't even part of collision because the default shape method didn't consider calculating collision for it.I see that you set all sub shapes to not-collidable. Here is the thing to keep in mind. Setting all sub collisions to not-collidable will do nothing. The not-collidable can only be applied to the top groups, not subgroups. Set particular subgroups to Ignore if you don't want them to be part of collision. To remove the not-collidable flag from all subgroups at once, simply select all subgroups of the cabin, right-click and select, SketchyPhysics > State > Collidable.
Even though the SP3.5.4 method for default shape is more profound, I will still revert this change to remain compatible with prior models.
Edit: Despite that, I just found a bad bug in default shape. I will fix it for SP3.5.5
-
I'll change the model accordingly and report briefly. I'm worried that you want to turn back more profound methods for older models.
-
Okay, fixed it! Now, the default shape should behave like in prior versions. I just adjusted the new method to act like an old one, instead of replacing with the original method from SP3.2. See how it goes!
-
Now it works better than ever before. Great! Convexhull2 and compound2 are very usefull shapes too. Thank you again!
-
-
Thanks Anton_s for the amazing amount of work you are doing!
A couple of notes:
@unknownuser said:
Fixed a bug where staticmesh inside a group would force simulation to crash. A staticmesh or compound within the group becomes default shape now.
This bug still exists for me using WinXP-SP3, SU 13.0.4812 and SP 3.5.5
@unknownuser said:
Added continuous collision checkbox option to the emitter. Enabling this will prevent emitted bodies from passing other bodies at high speeds. Now, bullets will collide if CC is enabled.
Where do I find the checkbox?
Additionally, a "make hole" function in SP would be amazing and time-saving!
-
@jumpjack said:
This bug still exists for me using WinXP-SP3, SU 13.0.4812 and SP 3.5.5
Can you upload a sample model?
@jumpjack said:
Where do I find the checkbox
Open the SketchyPhysics UI, and in the Emitter field, you'll find Emit Continuous Collision Mode checkbox.
@jumpjack said:
Additionally, a "make hole" function in SP would be amazing and time-saving!
Do you mean a hole in an object? Like, um, the bullet holes?
Advertisement