Doorways and the new Firestorm Physics View

[Chic Aeon]

I have one of those looooong bakes happening and since I already got some tea I thought I would share this --- mostly for new folks that might get pretty confused.

I don't know if this happens with ALL doorways, with just cube (analyzed) physics or what but it can be fairly disconcerting when you think your physics model isn't working.

 

Here's the deal.

After I uploaded a mesh entry wall (with doorway) I wen't into the build window and looked at the physics with the new "see the physics model" option in the Features menu (the little eye up by Physics Shape Type).  Well I at least THOUGHT a word that you can NOW say on TV. Why wasn't my physics model working.

 

Well the answer is, it WAS WORKING -- it just didn't look like it in the "see the physics" pane.  Here are the screenshots. I also uploaded the same mesh with high physics (just choosing that option in the uploader). It showed the same solid door but a slightly different shape.  The doorway WORKS just as it should.

So don't panic if this happens to you.  And my bake is done. Pasting in screenshots.

 

EDIT: The 10 land impact doorway is the one with the HIGH physics from the uploader  which shows you why you want to make your own physics models LOL. 

Edited February 19, 2018 by Chic Aeon

[Chic Aeon]

I just uploaded a VERY simple building with three doorways and all was as expected.  

This was not just a Beta Grid Occurrence  as the same shows up on the main grid.  Again this was analyzed physics.  

That's my report. So it may have been the complexity of the doorway?  No clue really. 

[Love Zhaoying]

Back in my day, we didn’t have this fancy “mesh”, or even “sculpties”. No by gum, we made our buildings out of PRIMS and we LIKED IT! They looked like they were designed by a 4-year old, but we didn’t care. And, we walked to our jobs uphill both ways in the snow. Naked. Because we spent all our money in SL, and we LIKED IT! A griefer could make our builds physical so they fell apart and we cried, and cried. But then we sucked it up and grew a pair, because we were TOUGH, we had GRIT and SPUNK!

[Chic Aeon]

I want to add more to this thread on physics as I had another issue today and it was on a VERY SIMPLE floor (so this isn't an issue with the complexity of the model).  

I uploaded twice again, once using high physics and once using cube physics. Both times it LOOKED as though the physics was wrong when in fact it was as it should be. Since it was a flat floor and I didn't turn it to upload I couldn't actually see what the uploader showed but I am guessing it is the "correct" look that you would expect, like with the door the other day.

They both showed the area closed, a little difference but basically the same --- BUT.

I check in the Develop Menu and it looked exactly the same as it did in Firestorms physic viewer. So this ISN'T a Firestorm issue and I am guessing that data gets pulled directly from the Develop >Show Metadata info. 

Lani stepped over the edge many times in many places to confirm that the physics were actually fine.  I did try planes physics and couldn't get that to work at all, but even though Aquila patiently showed me the ropes on that and I used it successfully a few times, it is not my norm so THAT might just have been me. My tests though had the physics staying convex hull even when changed to prim on the NON-ANALYZED physics. Likely someone has done this and can add some info here. Likely just me with the planes physics.

Here are some shots.  New shop floor -- hence "oversized" scale :D. 

 

 

Edited February 21, 2018 by Chic Aeon

[Charlotte Bartlett]

That's kind of frustating as I hoped to have quicker physics checks, I have to be honest I always revert to the LL viewer for anything with physics to upload - have you seen if the behaviour is consistent on that viewer too in the develop menu?  

[Aquila Kytori]

23 hours ago, Chic Aeon said:

Lani stepped over the edge many times in many places to confirm that the physics were actually fine.  I did try planes physics and couldn't get that to work at all, but even though Aquila patiently showed me the ropes on that and I used it successfully a few times, it is not my norm so THAT might just have been me. My tests though had the physics staying convex hull even when changed to prim on the NON-ANALYZED physics. Likely someone has done this and can add some info here.

 

A possible reason why you couldn't get the planes, non analyzed physics to work :

If a Non analyzed physics mesh is less than 0.5 meters thick along the X or Y or Z axis then even when set to Physics Shape Type: Prim it will act as idf it were Cobvexx Hull.

This is not a bug. This is expected behavior.

I am guessing that this is the case for both your "Doorway" in your first post and the "Floor" in your second.

To work around this an extra triangle can be added to both the visual mesh object and the physics mesh so that they both become equal to or a little more than 0.5m thick.

Or you can simply add an extra vertice to do the same thing. (Both visual and physics models).  Normally single unconnected vertices are ignored in the .dae file. But ......... if you make this extra vertice first in the list of vertices then it will not be ignored.

To do that, select the extra vertice,  open the Mesh menu > chose the Sort Elements option > then choose Selected.

 

I just did a test with 2 floor objects:

The original floor mesh is only 0.2m thick, not Analyzed an set to Physics Shape Type Prim. The second floor is using the extra vertice so that the mesh is now 0.6m thick. Upload and other settings are exactly the same as the original floor model.

This is how they look in the Firestorm viewer :

 

And this is how they look in the SL viewer:

 

The Sl viewer is showing them both the same even though the original is actually acting as convexhull.

It is the Firestorm viewer that is correctly showing the Physics as they really are.

 

 

[Chic Aeon]

11 hours ago, Aquila Kytori said:

A possible reason why you couldn't get the planes, non analyzed physics to work :

If a Non analyzed physics mesh is less than 0.5 meters thick along the X or Y or Z axis then even when set to Physics Shape Type: Prim it will act as idf it were Cobvexx Hull.

This is not a bug. This is expected behavior.

I am guessing that this is the case for both your "Doorway" in your first post and the "Floor" in your second.

OK. I get that and remember from long, long ago that is why I started using "cube" analyzed physics (also long ago).   That might explain the plains planes physics version not working at all.

 

Edit: Actually I did KNOW this but in the midst of all the other confusing "blueness" I simply forgot about it. That long discovery and discussion from way back when was one of the first I remember from this forum. And amazingly I understood it.  I am a bit puzzled why items less than 20 inches ish give this behavior as expected though. That is a REALLY thick floor or window casing. 

That is a very cool way to fool the uploader though. Seems like it would be better if we didn't have to go through all these extra steps. Perhaps that will get on the list of the great new fixes Linden Lab is planning for the future. Addressing land prices (notice lowering is not the verb LOL) is good. But solving inventory issues (apparently an insurmountable problem and a more reasonable uploader would be  plus in my book. 

 

BUT my original tests shown as noted were CUBE physics. I only tried the planes physics on the floor after the cube physics failed to look correct but worked fine a couple of times (and different viewers).

So maybe I am missing something?  Long day here so I may have :D. 

I just looked at my file and both door and floor are less than .5 as you suspected BUT I still don't see that explaining the issue with the cube physics (that did work just fine but didn't LOOK like it worked).

AND then there is the instance of the STAIRS which are quite large in all directions and one single mesh object.

Sooo ? 

I am still confused on the cube physics issue ohgreatwiseone :D.

Sure those pretty pictures will help someone!!!! 

 

 

Edited February 23, 2018 by Chic Aeon

[Whirly Fizzle]

5 hours ago, Aquila Kytori said:

The Sl viewer is showing them both the same even though the original is actually acting as convexhull.

It is the Firestorm viewer that is correctly showing the Physics as they really are.

Yes, Beq fixed this in the latest Firestorm release.
http://hg.phoenixviewer.com/phoenix-firestorm-lgpl/rev/87a852294d54

LL JIRA; BUG-134006 - Viewer code is not aligned to server code when calculating physics shape for thin objects.

 

[Anna Nova]

(off topic)  I want a course in making explanatory pictures like Aquila Kytori does.