Mesh collision physics after export or sim restart

Building using prims, sculpties and meshes. Texture creation techniques.
User avatar
Sherrie Melody
Posts: 273
Joined: Fri Mar 29, 2013 6:56 pm
Has thanked: 201 times
Been thanked: 159 times

Re: Mesh collision physics after export or sim restart

Post by Sherrie Melody »

Also, Zuza, what viewer and version of the viewer are you using?
User avatar
Zuza Ritt
Posts: 235
Joined: Mon Dec 24, 2012 10:25 pm
Has thanked: 205 times
Been thanked: 308 times
Contact:

Re: Mesh collision physics after export or sim restart

Post by Zuza Ritt »

Sherrie, that experiment I have tried before with highest and lowest LOD, and it doesnot matter. Even if some physics is calculated in upload Tab, there is some OTHER physics used after import, not the one which was uploaded. Maybe it is medium LOD.
I tried it in at least two versions of Firestorm and Singularity and I will not switch the viewer, because it is obviously not viewer related, this problem is known some time and not only for me.
It is obviously feature of OpenSim, to switch the mesh into Convex Hull after something like import, restart, maybe some other things.
I live among the trees. You are invited too:) Visit 21strom.
My Kitely Market. 21strom webpage.
User avatar
Sherrie Melody
Posts: 273
Joined: Fri Mar 29, 2013 6:56 pm
Has thanked: 201 times
Been thanked: 159 times

Re: Mesh collision physics after export or sim restart

Post by Sherrie Melody »

Zuza, I wanted to try the same experiment I suggested to you, to see how it worked for me.

Just for reference, I am using Firestorm 4.2.2 (29837).

I uploaded a mesh object, in the physics tab setting level of detail to "High"
It imported as phantom, so I deselected Phantom in the Edit>Object tab. The physics worked correctly.
I then exported the OAR from Kitely.
Then I imported the OAR back into Kitely.
The mesh object I'd uploaded retained its correct physics behavior.

Because the OAR import was successful for me, that means that, although it isn't working for you, it is at least possible to import OARs that retain correct physics behavior on mesh objects.

Please understand, I'm not suggesting you are doing anything wrong at all, it's just that with more details, it might help us to narrow down the problem that is occurring.

Can you tell me which particular version of Firestorm you are using?
User avatar
Sherrie Melody
Posts: 273
Joined: Fri Mar 29, 2013 6:56 pm
Has thanked: 201 times
Been thanked: 159 times

Re: Mesh collision physics after export or sim restart

Post by Sherrie Melody »

I wanted to try a couple more experiments. In the last experiment, my mesh object was a single item, a boat, 1 prim in kitely.
So, I create a 3 item mesh object (just a series of walls), 3 linked prims in kitely.
I uploaded the mesh object into Kitely twice, once setting the physics to "High" and a second time setting the physics to "Medium", both imported properly with physics working correctly.

I then exported the OAR from Kitely, then re-imported the OAR to Kitely. The mesh objects retained the correct physics behavior.

It would be great to get some other people's input on their experiences with mesh physics behavior after OAR export/import, including the exact procedures, viewer, viewer version, etc. they were using to upload mesh. It might help in finding the specific issue that some people have.
User avatar
Zuza Ritt
Posts: 235
Joined: Mon Dec 24, 2012 10:25 pm
Has thanked: 205 times
Been thanked: 308 times
Contact:

Re: Mesh collision physics after export or sim restart

Post by Zuza Ritt »

Sherrie, thank you very much for your experiments and help.
I agree, it has to be explore more.
I tried to reproduce behaviour which I described earlier with the same mesh as day before, but as only object in region, and it was "unsuccessful" - mesh have got proper physics after imported in OAR, and even after restart of OpenSim.
So now I am more confused:) It can be related to other mesh objects? It would be a nightmare for merchants...
I am preparing entire region with mesh landscapes, I would like to import it into Kitely and let people walk these landscapes... so I will continue in experimenting, it is crucial for big portion of my creations and for the future ones, because now I am doing mostly meshes....
I will write my observations, if I will find something more specific.
Thanks again;)
I live among the trees. You are invited too:) Visit 21strom.
My Kitely Market. 21strom webpage.
User avatar
Sherrie Melody
Posts: 273
Joined: Fri Mar 29, 2013 6:56 pm
Has thanked: 201 times
Been thanked: 159 times

Re: Mesh collision physics after export or sim restart

Post by Sherrie Melody »

You are welcome Zuza. OpenSim has supported mesh for a long time. However, OpenSim users have been slower to developing in mesh than users in SL. That means that in OpenSim, fewer people are reporting bugs to OpenSim and to their viewer developers, which means that focus on mesh issues is probably lower on the priorities. And, for you and me, it's really quite difficult to know if a problem is being caused by our viewer or the OpenSim. The best thing we can do, is when we discover a problem, to report it perhaps both to our viewer developers and the opensim developers as well, and report it precisely with detailed information about viewer version, opensim version, precise steps taken to produce the problem, etc.

Not necessarily directed at you Zuza, but for any of us in opensim environments, to get consistent results with mesh:
1. experiment.
2. document precisely the steps taken to get a successful result.
3. document precisely the viewer version and opensim version in which the successful result was produced, this includes where we upload the mesh and export OARs, and where we import OARs.
4. follow our documented process consistently.
4. if we change viewer, or change viewer version, repeat the experiment to make sure our process still works. If not, document a new process (if it exists) for getting successful results in the different viewer.
5. if we are working in a new version of opensim, repeat the experiment to make sure the process works. If not, document a new process (if it exists) for getting successful results in the different opensim version.

I am terrible about reporting bugs. :oops: I'll try to resolve now that I will do better in the future. For those of us working in mesh, it is probably extra important that we report bugs, because there aren't that many people working with it.

The opensim bug tracker is here: http://opensimulator.org/mantis/my_view_page.php

The firestorm viewer bug tracker is here: http://jira.phoenixviewer.com/browse/FIRE
User avatar
Zuza Ritt
Posts: 235
Joined: Mon Dec 24, 2012 10:25 pm
Has thanked: 205 times
Been thanked: 308 times
Contact:

Re: Mesh collision physics after export or sim restart

Post by Zuza Ritt »

Sherrie, thanks for the how to, I believe there will be more people with this problem...

I am back again with my problem today... and the worst thing is, I do not know WHAT to report... I just prepare about 20 mesh objects in one OAR, physics was perfect at my standalone, after importing OAR to Kitely all the object have something, I do not know what, visible physics by Rendering Metadata - Physics Shapes (in Firestorm) show something, which is not true, because "shape" of invisible block is different, probably like bounding box of all the object. Object looks like well set - root prim is Prim Type Shape, children are None. I just switch this ones again, in same values, root for Prim, children for None, technically, no changes...and voila, object is ok after this "change", I can enter it... so, is it bug? Is it some magic? Ghost on the sandbox?:)))
I am not sure WHAT to report:) Every try make different results... and no, I am not crazy:)

So now, I will try to write script, which does this magic, switch the value to the (same) values, when object is rezzed or after sim restart... hope something like this is possible:) And maybe it will "solve" the problem:) I will document this, we will see...
I live among the trees. You are invited too:) Visit 21strom.
My Kitely Market. 21strom webpage.
User avatar
Ilan Tochner
Posts: 6527
Joined: Sun Dec 23, 2012 8:44 am
Has thanked: 4990 times
Been thanked: 4473 times
Contact:

Re: Mesh collision physics after export or sim restart

Post by Ilan Tochner »

Hi Zuza,

It's best not to create workarounds but rather to create simple scenarios that consistently reproduce the problem so that we, or other OpenSim/viewer developers, will be able to find and fix the bug that is causing the problem. The clearer and simpler the scenario the quicker we'll be able to find the problem.
These users thanked the author Ilan Tochner for the post:
Dundridge Dreadlow
User avatar
Sherrie Melody
Posts: 273
Joined: Fri Mar 29, 2013 6:56 pm
Has thanked: 201 times
Been thanked: 159 times

Re: Mesh collision physics after export or sim restart

Post by Sherrie Melody »

Zuza Ritt wrote:Sherrie, thanks for the how to, I believe there will be more people with this problem...

I am back again with my problem today... and the worst thing is, I do not know WHAT to report... I just prepare about 20 mesh objects in one OAR, physics was perfect at my standalone, after importing OAR to Kitely all the object have something, I do not know what, visible physics by Rendering Metadata - Physics Shapes (in Firestorm) show something, which is not true, because "shape" of invisible block is different, probably like bounding box of all the object. Object looks like well set - root prim is Prim Type Shape, children are None. I just switch this ones again, in same values, root for Prim, children for None, technically, no changes...and voila, object is ok after this "change", I can enter it... so, is it bug? Is it some magic? Ghost on the sandbox?:)))
I am not sure WHAT to report:) Every try make different results... and no, I am not crazy:)

So now, I will try to write script, which does this magic, switch the value to the (same) values, when object is rezzed or after sim restart... hope something like this is possible:) And maybe it will "solve" the problem:) I will document this, we will see...
Zuza, I'm sure that by now, for the time being you have become tired of experimenting, especially considering your frustrations with today's experience. :cry: and if the scripting solution works, I will look forward to your describing it here.

If you were up to it, however, and wanted to try to track down the specific problem:
First, I think I would try uploading those 20 mesh objects to Kitely, and export the OAR from Kitely instead. I think Kitely has done some things to improve OAR export.

If that is unsuccessful as well, I would try a series of upload/import/export procedures. First with one mesh object, next add another mesh object, and so on. Each time check if the physics are good, and if so, continue to add another mesh object, export OAR, import OAR. If you are doing this in your offline instance, each time you add a new mesh object and export OAR and import the OAR, you can check the console to see if there are any errors generated on the OAR import. If mesh physics are messed up on an object, you should see an error in the console. You can then report a bug, providing details about your upload/export OAR/import OAR process (including exact inputs you added in specific fields in each tab), your viewer and version, and opensim version, and also include the full text of any errors generated in the console.

---------------------------------------------------------------
(I'm not sure if this will help, but is what I would do) Each time you add a mesh object, if you find that the physics are messed up on that object, remove the object before you add a new mesh object in your next trial. The goal would be to see if there is a specific mesh object, or type of mesh object that causes the fail.

---------------------------------------------------------------

By the way, your halloween setup (http://www.kitely.com/market/product/27 ... -landscape) looks totally cool! If you had an export version, I'd buy it in a heartbeat :D

Edit: I totally forgot to add a bit of stuff above, the part between the dashed lines.
These users thanked the author Sherrie Melody for the post:
Ilan Tochner
User avatar
Sherrie Melody
Posts: 273
Joined: Fri Mar 29, 2013 6:56 pm
Has thanked: 201 times
Been thanked: 159 times

Re: Mesh collision physics after export or sim restart

Post by Sherrie Melody »

In the interest of sharing experiences on mesh import, thought I would add this. Ozwell had two sets of mesh trees in the marketplace that I really liked, and he has C/M + export permissions, so I bought them today. I rezzed them in world and they looked great. They are all multi-prim linked items. A total of 8 objects. They also had physics set so if you walk into the tree trunk, you get stopped, so you have to walk around the tree.

I figured since they were export and had physics enabled, it would be a great opportunity to experiment with more complex mesh items in the export OAR/import OAR thing. So, I exported the OAR from Kitely, and imported the OAR back into kitely, and the trees retained their physics properties correctly. The trees all have between approximately 1.3 thousand triangles and 2.3 thousand triangles. (this information is available from the Develop>Show Info>Show Render Info menu item).
Post Reply