Disappearing textures and structures

[selden]

Harley,

Sorry, I've lost track of which problem I've discussed with which victim...

Have you tried viewing your model using Celestia v1.3.1 or v1.3.0?

There were bugs introduced in v1.3.2 as as side effect of a substantial overhaul of the model display code. Most of them got fixed before the final release, but some are still there.

[Beowulf01]

I'm running Version 1.3.2

[Beowulf01]

I tried a new model this afternoon, i get the cell structure but the individual solar cells disappear when i view them from"behind".

Also, id like to ask about alignment.
I need the cells to be pointed at the SUN. i now i can set the rotation to mate its orbit, but the initial appearance in orbit needs to be facing the sun. how do i do that?

[Beowulf01]

ok i uploaded the new pics.
i did a bit of work and it looks like i got this puppy nearly done.!!
I just need to figure out why the cells dissapper when viewed from behind.
also This model is REALLY slowing down my machinf.
I got 1Gig of PC3200 DDR memeory and a nvidia 4600 Ti card.....
might have to reduce the Polys?

[selden]

Translating from 3ds to cmod will give a 1.5-2x speedup.

# CMOD Tools: 3dstocmod, cmodfix
(require Celestia v1.3.2 or later)

* Windows binaries: 3DS to CMOD, CMODfix: http://www.spacesim.net/tools/cmodtools-win32-1.0.0.zip
* Documentation:
http://www.celestiaproject.net/~claurel/celest ... readme.txt
* Sourcecode: http://cvs.sourceforge.net/viewcvs.py/c ... ools/cmod/

What's the current poly count? I haven't seen any speed reduction when displaying models containing about 300,000 points when using a Ti4200.

One major cause of slowdowns is the test for obscuring objects. How much does the perforance improve if you turn off galaxies and/or stars?

[Toti]

My intent was to show how surface normals and ordering of vertices in a model can affect the visibility of a surface.

I don't know if this will be useful, but here it goes:

This effect has something to do with the so called "right hand rule" in linear algebra.
Place your right hand so its palm touches the 0-2 edge. The thumb will point to you.

Code: Select all

1--2
| / 
|/   
0


Imagine that you are pushing the 2 so it reaches the 1. This gives you some kind of order: the triangle defined as 0 2 1 has a normal (the thumb) "directed to you". The same is true for 2 1 0 and 1 0 2 (Note that we are describing a counterclockwise path)

If you repeat this (always using your right hand) for 0 1 2, you get a normal pointing in the opposite direction.
The same is valid for 1 2 0 and 2 0 1 (clockwise)

In Selden's example, if we watch the trilist square from "above" (the y axis):

Code: Select all

x axis
^
|

3__2
| /|
|/ |
0__1         -> z axis


The paths 0 1 2 and 2 3 0 give a positive normal (along the +y axis)
But 0 3 2 and 2 1 0 gives a negative normal.

This is done in a per face basis: it's useful to throw a measure of where the face is pointing in its own planar space, independently of how it's oriented in the 3D world.

[Beowulf01]

Uh......................
damm I'm dumb,
I think this is a bit over my head...

but i think I'm on my way of getting the model sorted out.
thanks toti.
selden, that conversion>.........? how doe sit work...
I guess i need to spend some serious time to figure it out..
sorry for all the trouble.

[Toti]

Harley,

Sorry, I was responding to Selden, not trying to address your particular problem. My post above is not directly related with the thread's subject. I edited it so this is more clear now.

Concerning the invisible faces:
Is your 3d model "double sided"? (ie: does each face has two sides)? I don't know if Celestia makes any difference, but you can check this. Rhino 3d should have some button to toggle double sides on/off.
If not, try a "point normals outside" kind of command.

If this still doesn't work, use the 3dsconv and cmodfix tools (see Selden's post above):


First convert from .3ds to .cmod:

Code: Select all

3dsconv your_3ds_file.3ds > any_name.cmod

Then fix and recompute normals:

Code: Select all

cmodfix -u -w -n -s 0 any_name.cmod  any_name_out.cmod

I never used the tool in this way, so I'm not sure if this will work.
Hope this helps.

[Beowulf01]

Thank god the site is back up!
OK I think i got this puppy whipped!!
I got the normals flipped and reduced the poly count... i had over 300000 polys ..............

I got a few things that concern me, such as corruption of he model when it is loaded in celestria, but I'm working around it.

Thanks.

[Guest]

These sound like the sort of models I would really like to have!
I'm looking at making a few more orbital habitats of my own; the normals are a bit recalcitrant in the ones Ive made so far as well, so perhaps I'll try converting to Cmod myself...
here is that inpterstellar ship again, with its ice shield intact, about to leave Earth orbit...
http://www.siderealgames.com/uploads/earth_and_ship.jpg

[eburacum45]

logs in to claim previous post

[Beowulf01]

looks liek you have a good start!
well done.

I took a break from working on my models.
Ill get back to it here tommorow... HL2 and Counterstrike source has been distracting me.