SF models here

[Cham]

I was experimenting a lot with the new specular reflection code (CVS 1.5.0), and I've edited many models (real spacecraft and SF models too). I'll publish some of the SF models here, starting with this one (real spacecraft are presented in another thread) :



If you want, you can download the model from the link below (the SSC is temporary and used only to show the model with its glowing lights). Enjoy !

http://nho.ohn.free.fr/celestia/Cham/Enterprise.zip (508 KB zip file)

[Cham]

Here's another one :



Enjoy !

http://nho.ohn.free.fr/celestia/Cham/Vador.zip (1.2 MB zip file)

EDIT : I forgot to edit the SSC. You probably don't have the planets around Altair, so just use your favorite SSC ...

O well, and I forgot to say those models should be used with Celestia CVS (aka 1.5.0), to enjoy the new specular reflections.

[Cham]

And another one :



http://nho.ohn.free.fr/celestia/Cham/lost.zip (504 KB zip file)

[ElChristou]

Tx for the work!
I'm not a big user of SF models but from time to time...

[chris]

Is that emissive texture or material on the Enterprise model? Nice to see that feature getting some use. It's especially useful when a spacecraft slips into the shadow of a planet.

--Chris

[Cham]

Is that emissive texture or material on the Enterprise model? Nice to see that feature getting some use. It's especially useful when a spacecraft slips into the shadow of a planet.

--Chris

Actually, it's Jestr's trick. The model was separated into two parts, one for the lights with an emissive command in the SSC. I want to use the new code features, but currently I'm unable to make them to work.

Chris, could you explain what we are supposed to do in the modeler, so Celestia could recognise the emissive textures/material ? And what about the normalMap ? I wasn't able to use that yet.

[buggs_moran]

Heh, as soon as I saw the "Lost in Space" model, the theme song started repeating in my head...

[Captain Nephilim]

The Enterprise looks great! Thanks.

[danielj]

Of course I can be wrong,but I think Nephilim is a Xenosaga character or a spaceship(the trilogy now in PS2).Is it a homage?

The Enterprise looks great! Thanks.

[selden]

Chris, could you explain what we are supposed to do in the modeler, so Celestia could recognise the emissive textures/material ? And what about the normalMap ? I wasn't able to use that yet.

Well, I'm not Chris, and I don't use NormalMaps, but I got emissive to work in a model.



If you don't need an emissive texture map, you just have to add the emissive color declaration to the appropriate material definition. This is what was emitted by my Anim8or cmod export script for the "glass" windows in the picture above. (I'm still working on it.)

Code: Select all

material # 58

# diffuse  0.788 0.957 0.98
#  diffusepower 1     # Kd
  diffuse 0.788 0.957 0.98
  opacity 0.8
  specular  0.576 0.576 0.576
#  specularpower 0.8     # Ks
  specpower 204     # Ks*255
#  emissive  0.992 0.945 0.627
#  emissivepower 0.5     # Ke
  emissive 0.496 0.473 0.314 # product
end_material


[cartrite]

Cham wrote:

And what about the normalMap ? I wasn't able to use that yet.

I'm not Chris either, but tangents must be generated for the model. cmodfix -t a.cmod b.cmod. I'm not sure how to generate them for 3ds model though.

[Dollan]

Hey Selden... Is that Orion model available somewhere? Or is it a copyright one? Thinking of something for the ArcBuilder setting, here....

...John...

[selden]

I'm working on it! I'm working on it!
Patience!

Pictures of some of my earlier work on the model are on the Web at http://www.lepp.cornell.edu/~seb/celest ... index.html

My primary emphasis at the moment is to try to reduce the resources it requires. It's running at only 15 fps on my home system (3.4GHz P4 + GF6600GT) and half that on the system at work (2.4GHz P4 + FX5200).

I've got some more meshes that can be optimized, but I fear a large part of the problem is that it consists of a dozen separate models. Pieces will get left behind during the expedition to Mars.

[chris]

I'm working on it! I'm working on it!
Patience!

Pictures of some of my earlier work on the model are on the Web at http://www.lepp.cornell.edu/~seb/celest ... index.html

My primary emphasis at the moment is to try to reduce the resources it requires. It's running at only 15 fps on my home system (3.4GHz P4 + GF6600GT) and half that on the system at work (2.4GHz P4 + FX5200).

I've got some more meshes that can be optimized, but I fear a large part of the problem is that it consists of a dozen separate models. Pieces will get left behind during the expedition to Mars.

The fact that there are twelve separate models shouldn't cause any problems (especially not in the current CVS version.) How many triangles are in the model? Does it run substantially faster in one of the non-OGL2 render paths? I know that there can be some slowdown when using a lot of materials. An option to condense identical material definitions to a single material would be a very useful feature for cmodfix.

--Chris

[selden]

Chris,

The largest model has about 300 meshes, so there are about 300 material definitions. Merging all the meshes into a single mesh reduces it to 14 materials, so eliminating duplicates would help a lot. Unfortunately, I can't merge all the meshes until Anim8or's export surface normal problem is fixed in the next beta. At the moment I need to use cmodfix to regenerate the normals, but it can't handle more than one material per mesh.

There seems to be about a 40% speedup when using the Basic render path compared to using the OGL2 render path with an FX5200 graphics card.

At one time I thought I had it up to 60fps on my home system, but I haven't been able to reproduce it. Maybe that was in the Basic path, I'm not sure.


Dollan,

Don't forget that the 10-meter version of Orion is only good for interplanetary travel at relatively slow speeds. Its isp is only about 1800. A much larger plate is needed in order to get the isp values up to where external plasma pulse drives can almost be used for interstellar travel.

[ElChristou]

...My primary emphasis at the moment is to try to reduce the resources it requires. It's running at only 15 fps on my home system (3.4GHz P4 + GF6600GT) and half that on the system at work (2.4GHz P4 + FX5200).

I've got some more meshes that can be optimized, but I fear a large part of the problem is that it consists of a dozen separate models. Pieces will get left behind during the expedition to Mars.

How many poly do you have actually?

[selden]

ElChristou,

They aren't all that bad, I think. The largest model has

Code: Select all

# 284 materials (+ 0 unused); 106608 vertices, 128629 triangles
# 259 Shapes    (+ 25 with no vertices; max: 6128, avg: 411,
#                         max Triangles/Shape: 5424, avg: 496)


[ElChristou]

Seems reasonable... and how many poly for the whole stuff?

[chris]

Chris,

The largest model has about 300 meshes, so there are about 300 material definitions. Merging all the meshes into a single mesh reduces it to 14 materials, so eliminating duplicates would help a lot. Unfortunately, I can't merge all the meshes until Anim8or's export surface normal problem is fixed in the next beta. At the moment I need to use cmodfix to regenerate the normals, but it can't handle more than one material per mesh.

There seems to be about a 40% speedup when using the Basic render path compared to using the OGL2 render path with an FX5200 graphics card.

At one time I thought I had it up to 60fps on my home system, but I haven't been able to reproduce it. Maybe that was in the Basic path, I'm not sure.

The 6600 should be able to handle the triangle count, but I'm not surprised that the 5200 slows down with such a large amount of geometry.

If the 6600 can display the model at 60fps without OpenGL 2.0, that would support my theory that the overhead of switching materials is to blame. That would actually be a good thing, because uniquifying materials is easy to do. It could even be performed with a relatively low cost at model load time in Celestia so that even existing models could benefit.

--Chris

[selden]

ElChristou,

I actually haven't done that arithmetic before now. It looks like the other pieces will add about 50%: ~57K vertices, ~85K triangles.