Back to Celestia, having changed nothing on the model, today I got the robotic arm ...
I also once got a scene with the old component and the upper solar panels (forgot to save that one in a picture

Marco
selden wrote:A quick check of surface normals can be done in Anim8or by going into the "point editor" and selecting all of the surfaces. If the surfaces are drawn blue, their normals are pointing in the wrong direction. If they're yellow, they're OK.
:D![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
Phoenix has landed on the Martian surface in Celestia.
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
But there are still a few issues on the model.
a) The texture of the upperdeck is black (texture2.jpg) instead of orange/gold (texture5.jpg)
b) The solar panels as seen from below must be black textured (texture2.jpg) instead of disorted solar panel pictures from (texture4.jpg)
c) The white space in the 2 towers on the upper deck must be transparent. I learned something in this post about Alpha codes in the according .PNG texture file, but apparently there is something incorrect now on this point.
d) The back side of the left (big) tower on the upper deck is textured with 3 stretched white areas, where the 2 front sides (it's a triangle) are correctly textured with 6 white areas. In the texture files there is only 1 .PNG file available, having this texture.
Fenerit wrote:But there are still a few issues on the model.
a) The texture of the upperdeck is black (texture2.jpg) instead of orange/gold (texture5.jpg)
b) The solar panels as seen from below must be black textured (texture2.jpg) instead of disorted solar panel pictures from (texture4.jpg)
c) The white space in the 2 towers on the upper deck must be transparent. I learned something in this post about Alpha codes in the according .PNG texture file, but apparently there is something incorrect now on this point.
d) The back side of the left (big) tower on the upper deck is textured with 3 stretched white areas, where the 2 front sides (it's a triangle) are correctly textured with 6 white areas. In the texture files there is only 1 .PNG file available, having this texture.
a), b), d) probably the same mesh has been mapped as double-sides (one side one texture) when the .dae model has been built. 3DS doesn't handle the double-side UV map, so the CMOD repeat the setting (dae format is more advanced than 3DS);
Fenerit wrote:c) For having the transparency, here you must take the ORIGINAL .png texture, open it into a graphic program and see whether the image has trasparent parts (your new is now possibly opaque). If positive, after resizeing and saving AS LAYER (32 bit), you must assign to the relevant CMOD mesh a bit of transparency, neither the full opacity nor the full transparency (e.a. an opacity of 0.3), save the CMOD model and reload Celestia. If all goes right, it's matter of taste establish how many transparency you need.
Marked with the red arrow, you can see a wrong texture on the back side of the big tower on the upperdeck. This texture need to be the same as the other 2 sides of the tower. So this error is due to a bug in the Collada exporter of the free Sketchup 7 program.
In Blender, you can easily select this part and delete it. I thought: "better no texture than a wrong texture, let's give it a try".
Then export it again as a .3ds, convert to CMOD, adjust all opacity values and there it was in Celestia. ALL the above problems were gone and there were even more components on the upperdeck now that were drawn much better than before. The only minor thing left now, indeed the backside of the tower is not present anymore, I think it is a minor issue for this v1.0, also because it is not particulary disturbing (my opinion).
It seems to be a kind of sequence error in the .dae file between meshes and materials, because when this (wrong) component is deleted, all resulting components and materials look like to shift a position, resulting in all other errors mentioned above, automatically fixed at once without handling them.
Of course, changing ALL opacity values in CMOD with Notepad++ was not that difficult, but finding the relevant CMOD meshes and changing a bit there is no go for me yet.