Emissive seamless strange behaviour (bugs?)

[Fenerit]

In building my Mars's interior, I've found this result, no matter whether in .3ds or in .cmod format. When I test the 3D model in Celestia 1.5.0 pre4, I obtain this seam (sorry for the small pictures but I'm still unable to send more consistent files):



Rather, if I assign the directive "emissive true" in the .ssc the seam desappear:



But the most strange thing is that the sphere model (8k vertex) is the same as Mercury, which doesn't show seam at all even without the emissive directive:



I've tried several Mars'map in order to see if was the texture and even to raise the spher's polygons, but it's not so. (obviously my modeller doesn't show seam whatever map I use or vertex count). Anyone else has found this about personal models?

[Fenerit]

Problem solved! Was the model's orientation in my 3d program. In order to have the sphere's cut in front of the screen in the 3d view, the model may be turn of 90?°. Once finished I may set the original position and add the orientation strings in the .ssc. I've caught this solution because I'm being to test the model "as it is", and I've noticed that with an ambient light of 0.85 to 1come from the Lua Tools, the seam vanished. Nevertheless it's a strange affair, because if were relate with the shadows on 3d model, it should always happen, no matter which is the orientation; instead, as long as I've toggle around the planet with the proper orientation against the light, there aren't seams.

[rthorvald]

If i remember correctly, Cham discovered you need a smoothing angle of 90 when generating the CMOD to get rid of seams.

- rthorvald

[Fenerit]

Thanks Runar for the suggest. I had a problem with the poles of my break model (Mercury showed pseudo-elevation at poles) but I've fixed it with a smooth angle of 80?°, while Mars is good as it is (even this it's strange too).

The problem it's another again: the seam vanish with the shutdown of the computer and susseguent reboot. Just now, that I go ahead with the modelling and testing again with the bad orientation, the seam is no longer present both in .3ds and .cmod format! So, I think of confirm what I've said here:

http://celestiaproject.net/forum/viewtopic.php?t=11728
(Cartrite's thread concerning the 3dMars and Cmodtool);

at least for my graphic card (440 mx 128 with the latest drivers) the seams can be no matter esclusively of cmodtool, but even of OGL routines.
Note: often I obtain the seam after having loaded certain OpenGl screen savers of math formulas in motion.