FreeGameDev

I've begun to convert Star Track. The attached files are the updated .blend, and all of the changed resources. The exported track itself is not in either of the archives for reasons I will explain. There are no items on the track yet.

The reason I'm posting this work-in-progress is because I ran into an issue. I have all of the planets set to export as separate models so that I can tell the physics to ignore them. When I export the track, however, all of the planets spontaniously rotate. If you take the blend file and export it while watching Saturn's ring, the problem is illustrated well. Can somebody tell me why this occurs so I can fix the issue (and prevent it in any further work)? I use blender 2.49b.

And no, I'd prefer not to leave the planets in the main track model, because then I can't animate them (if I ever find the time to do so).

Thanks

[EDIT] Ctrl+A fixed the issue Hopefully I'll be able to have the rest of the track done by the end of the day now.

sj04736 {l Wrote}:When I export the track, however, all of the planets spontaniously rotate.

Just to make sure I understand you correctly: the planets rotate during the actual export (and not while running in stk)?

If you take the blend file and export it while watching Saturn's ring, the problem is illustrated well. Can somebody tell me why this occurs so I can fix the issue (and prevent it in any further work)? I use blender 2.49b.

I can't reproduce the problem. If you see something rotating during the export, it means there must be some skeletal animation happening (since IPOs are exported 'invisible' during the export). Can you check if there are any armatures defined?

It's also worth (just in case) to check the scene.xml file - look for any lines like:

{l Code}: {l Select All Code}

   <curve channel="LocX" interpolation="bezier" extend="cyclic">
      <p c="1.000000 34.090733" h1="-1.311990 30.471663" h2="3.311990 37.709801"/>
...


But those lines indicate IPO animations.

And no, I'd prefer not to leave the planets in the main track model, because then I can't animate them (if I ever find the time to do so).

Two comments: the planets appear to have a rather high number of vertices (around 1000 each) - it might be worth checking if the same visual effect could be done with lower resolution spheres. Since the actual track has only 2000 sphere, the 11 planets make the track 5 times as big. Also it appears that the driveline is modelled with a certain 'thickness', i.e. a kind of very flat torus. Do you know thy this is done? To avoid backface culling and the track then being invisible when loo0king up? If this is the case, wouldn't it be better if we supporte deactivating backface culling for certain materials or objects?

Cheers,
Joerg

hiker {l Wrote}:Just to make sure I understand you correctly: the planets rotate during the actual export (and not while running in stk)?

After some testing, I discovered that on export, it returns all objects' rotations to 0. Now that I know how to save the object's rotation (with Ctrl+A) that's no longer a problem.

Two comments: the planets appear to have a rather high number of vertices (around 1000 each) - it might be worth checking if the same visual effect could be done with lower resolution spheres. Since the actual track has only 2000 sphere, the 11 planets make the track 5 times as big.

I'm sure I could decrease the polygon count on the spheres without losing any quality.

Also it appears that the driveline is modelled with a certain 'thickness', i.e. a kind of very flat torus. Do you know thy this is done? To avoid backface culling and the track then being invisible when loo0king up? If this is the case, wouldn't it be better if we supporte deactivating backface culling for certain materials or objects?

I think you're correct here. IIRC, Gown's Bow (the parent track) had that problem where you could not see the track from below, and the person who modified it to create Star Track used this solution, rather than making the faces double-sided (which was possible with pLib, not sure about irrlicht).

Star Track is now on-par with the 0.6 version (no upgrades yet) [EDIT:] I did decrease the polycount on the spheres by a factor of 0.5 (or 0.25 for the smaller ones)

The only issue now is the disorienting rotating skydome

I didn't include 'Ethereal_Spectrum.ogg' in the archive because it's already in the repository.

Anyways, this is all I'll be able to do this week. I might be able to do some more next week-end.

hiker {l Wrote}:the planets appear to have a rather high number of vertices (around 1000 each) - it might be worth checking if the same visual effect could be done with lower resolution spheres.

Does STK [plan to] support LODs?

Thanks Stephen!
I added the track to STK and to the media repo. I also fixed the skycube

charlie {l Wrote}:

hiker {l Wrote}:the planets appear to have a rather high number of vertices (around 1000 each) - it might be worth checking if the same visual effect could be done with lower resolution spheres.

Does STK [plan to] support LODs?

Later perhaps, atm there is no 'native' support for LOD in irrlicht (except for terrains using height maps, which we are not using). Will be useful for karts (though there is of course the start, during which you will usually see all karts from close anyway, so LOD might not help that much).

I am not sure if we have to use this for tracks - I haven't had the time to look at performance yet. But in the example here I'd say that it is really not necessary to have spheres with that many vertices (while I think it wouldn't be much of a performance bottleneck, since most of them will be culled anyway, it just makes things unnecessary big, and slow e.g. at loading).

Cheers,
Joerg