Problem in dgWorldDynamicUpdate::BuildClusters

Report any bugs here and we'll post fixes

Moderators: Sascha Willems, Thomas

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby Julio Jerez » Thu Nov 15, 2018 6:52 pm

the compound fix in in the way. 8)
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby misho » Fri Nov 16, 2018 2:59 pm

Julio Jerez wrote:the compound fix in in the way. 8)


Which, if I understand correctly, will fix vertex cloud Debug/Release problem I'm having? :wink:
misho
 
Posts: 396
Joined: Tue May 04, 2010 10:13 am

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby Julio Jerez » Mon Dec 03, 2018 2:07 pm

Bird wrote:Yes, except for the compounds, which are crashing here, ...

I think the compound crash is fixed now.
if you sync an try again, it should be working now.

@misho, can you try again, I am not sure if this will fix you bug, because this was a contact generation bug, but anyway you can try now and let see what else is wrong.
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby misho » Thu Dec 06, 2018 7:38 pm

Julio Jerez wrote:@misho, can you try again, I am not sure if this will fix you bug, because this was a contact generation bug, but anyway you can try now and let see what else is wrong.


Hi Julio - I have just compiled with the latest - unfortunately I am still seeing differences in behavior in Debug and Release versions... as per my post related to vertex cloud collisions. Something changed, because there are now collision objects where previously there were none (in Release), but there are still problems.
misho
 
Posts: 396
Joined: Tue May 04, 2010 10:13 am

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby Julio Jerez » Thu Dec 06, 2018 8:02 pm

yes I fix a bug that when creation island the there was an assumption that if the one of the two bodies lined by a joint was static, then that body would be the second, however that's no really necessary and if some one made a joint like that, the inland would be wrong.
this should be an improvement,
one the different between release and debug, I do not know what could that be.
are you saying this happen in the NewtonMesh tool? or in the physics engine.

if this is the case can you export you mesh as an FBX file, maybe we can load it in the sand box
there is a tool now that can convert fbx file to ngd, if you can save you mesh as fbx they we can see if the bug can be reproduce in the sand box.
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby misho » Thu Dec 06, 2018 8:26 pm

Julio Jerez wrote:are you saying this happen in the NewtonMesh tool? or in the physics engine.

if this is the case can you export you mesh as an FBX file, maybe we can load it in the sand box
there is a tool now that can convert fbx file to ngd, if you can save you mesh as fbx they we can see if the bug can be reproduce in the sand box.


Hi Julio,

Since Debug works OK, and Release does not, I believe this happens in the physics engine. I do not use NewtonMesh tool. I have a file exporter from 3DS MAX which exports to DirectX .x format (an ASCII text file). From there, I have a loader that loads the mesh vertices from the .x file and creates vertex cloud from it. Therefore, it would be quite a pain to re-create the precise scene - my scene has about 10 objects attached into an assembly (rocket). In addition, there are another 20 or so objects (Newton Bodies) that are attached to the rocket parts that have their collision set to NULL - they are used as 3D rocket flame objects from various rocket engines and thrusters - so, they are 3D "plasma" effects which have no collisions.

I can certainly show you a screenshot of the difference between Debug and Release, if that will help...
misho
 
Posts: 396
Joined: Tue May 04, 2010 10:13 am

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby Julio Jerez » Thu Dec 06, 2018 8:46 pm

My guess is that the bug is in the part the get the vertex cloud, My guess is that there has to be some error that produce different vertex cloud in debug than in release.
what does the debug display say?

can you make a test build the link to the dll and send to me so that I can check out?
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby misho » Thu Dec 06, 2018 9:44 pm

Julio Jerez wrote:what does the debug display say?


I think you are misunderstanding how I'm using Newton. I am not using demoSandbox at all. I created a DLL client that uses Microsoft Flight Simulator X (FSX) as a server. My DLL uses MS's SimConnect API to "take over" objects in FSX and put them under the Newton physics model. My environment is FSX, not demoSandbox. So, I don't have "debug display", if what you meant was debug display within demoSandbox. I do have a console window where I can output any value needed.

Julio Jerez wrote:can you make a test build the link to the dll and send to me so that I can check out?


As per above, you probably thought that I have a dll that links in some way to the demoSandbox. I can certainly send you my DLL, but you will have to obtain and install FSX, and then properly install my project with all its content: 3D models, instrumentations, sounds, scripts, etc. (I don't have an auto installer yet, but I was going to implement one fairly soon because I'd like to release a demo of my simulator sometimes hopefully before Christmas).

To better illustrate, here is a screenshot of what I observe:

In the top frame, I remove a rocket component "A".
As a result, the part of the rocket that was attached to the top of it falls back and tilts towards the tower structure, and then it hits it and rests, leaning on it. The assembly, as shown, consists of 6 Newton bodies, including 3 panels marked "B", all attached using the "hard" hinges.

The middle frame shows how this happens in Debug version... all looks correct.

Image

The bottom frame shows how this happens in Release version. Notice the panels "B" now "dangling" (but still connected to the main assembly). Also - notice the gap "C" - there is a collision box here but it seems misplaced by a small amount. Before this latest build, this collision was not "registered" in Release build, and the rocket part would fall all the way down.

I hope this helps better illustrating what I'm seeing.

I can share anything you need from my project - I can send you the code I'm using to create vertex clouds if you want. Just let me know what you need...
misho
 
Posts: 396
Joined: Tue May 04, 2010 10:13 am

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby Julio Jerez » Thu Dec 06, 2018 10:36 pm

I know you are not using the sand box, that's not what I meant.
But anyway, can you just recreate that part of the rockect in the sand box, otherwise I do not know how to work on that. As far as. I know there is not difference between debug and release
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby misho » Thu Dec 06, 2018 11:26 pm

Ok, I'll try.

I have already created a similar scene in demoSandbox a few weeks ago, and I did not get the same problems. I guess I have to get a bit more closer to what I have. There are many factors in my setup - for example, I'm on the surface of the Earth, so I am ~6370000 meters away from the origin, and I'm at Florida latitude/longitude, so the rocket is tilted (of course, with gravity in the opposite direction towards center).

Then, I have to use the same loader and files to create my vertex clouds. It will take me a bit of time but I'll put it together.
misho
 
Posts: 396
Joined: Tue May 04, 2010 10:13 am

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby Julio Jerez » Thu Dec 06, 2018 11:37 pm

this double presition I assume.
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby misho » Thu Dec 06, 2018 11:42 pm

Yes, it is!
misho
 
Posts: 396
Joined: Tue May 04, 2010 10:13 am

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby Julio Jerez » Fri Dec 07, 2018 12:04 am

Here is a question, how do people get FSX?
Wouldn't it be eassier if I get a copy and you go me a copy of you plugin?
Unless it is a ridiculous expensive application.
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby misho » Fri Dec 07, 2018 12:44 am

Most definitely you can! It is sold on Steam, here is the link, and it is $28 (that may be Canadian price showing for me), definitely not expensive (I've seen it on sale for $12, there could be one coming for the holiday season).

After you install it, I can send you an archive of my project (dll and the content), and here is a link on how to manually install my add-on, although I may have to change a few things - this is for a year-old demo.

I guess you'll be able to run Newton libraries in debug mode while FSX + FSXSpacePort.dll are running, and trace what's going on from there? Or - if you need my whole source code VS project/solution, that's no problem either...
misho
 
Posts: 396
Joined: Tue May 04, 2010 10:13 am

Re: Problem in dgWorldDynamicUpdate::BuildClusters

Postby Julio Jerez » Fri Dec 07, 2018 3:19 pm

I will buy tomorrow, tell me this is the same Microsoft simulation that you can buy anywhere like in best buy of frys?
Julio Jerez
Moderator
Moderator
 
Posts: 10485
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

PreviousNext

Return to Bugs and Fixes

Who is online

Users browsing this forum: No registered users and 1 guest