Celestia Forums

25143 Itokawa's detailed shape models (Text,STL,TRI) were released.
http://hayabusa.sci.isas.jaxa.jp/

This is extremelly interesting, especially since I'm able to open the files and make a 3ds model of the asteroid. Here it is in my modeler :



However, this one has almost 800 000 triangles. This isn't a problem on my machine, but I'll have to slice it in several parts since the .3ds format can't accept pieces with more than 65 000 triangles. I asked Dirkpitt to add an option to its CMODTool utility, to convert .obj to CMOD too. He agreed with this, but he's not here anymore since a pretty long period of time now.

The smallest model has less than 50 000 triangles, but it's crude :



I can easily export that one to 3ds.

EDIT : I don't know what's happening. I'm yet unable to export the smallest one to 3ds. The modeler says it has more than 65000 triangle, while the info box says it has 49152 triangles !?

Hmmm, the model format is weird. Each triangle appears to be "separated" from the others. This makes thing more complicated for exportation.

Cham wrote:Hmmm, the model format is weird. Each triangle appears to be "separated" from the others. This makes thing more complicated for exportation.

I can convert this to a cmod after work tonight. I'll hack together a little tool to translate it to cmod, then run it through cmodfix to weld the triangles into a mesh.

--Chris

chris wrote:

Cham wrote:Hmmm, the model format is weird. Each triangle appears to be "separated" from the others. This makes thing more complicated for exportation.

I can convert this to a cmod after work tonight. I'll hack together a little tool to translate it to cmod, then run it through cmodfix to weld the triangles into a mesh.

--Chris

Thanks Chris. I'm unable to export the mesh. My modeler appears to be confused with the triangles. I tried to slice it in two, but it didn''t worked for unknown reasons. I'm even unable to smooth it.

I've converted the Itokawa shape models into binary cmods. I haven't yet processed the highest resolution one (over three million triangles!), but the others are done. Here's the 786,432 triangle model in action:



You can download them from http://www.celestiaproject.net/~claurel/celestia/asteroids

itokawa0.cmod - 49,152 faces - 4M
itokawa1.cmod - 196,608 faces - 16M
itokawa2.cmod - 786,432 faces - 64M

Please only download the largest one if you have respectable graphics hardware and actually intend to use it. Otherwise, you're just wasting shatters.net bandwidth.

I will soon try and incorporate these models into a complete Hayabusa add-on.

--Chris

VERY nice ! Thanks Chris. I'll download the largest one tonight. I like very much this "peanut". I was thrilled when the japanese scientists published the pictures of the real thing, last year. Now, we have it with a nice resolution in Celestia, isn't that great !?

I have the 3 million triangle version converted to cmod now. I haven't put it on shatters.net, but if anyone wants it, send me a personal message on the forum and I'll make it available. It's 256 megabytes, and even my GeForce 8800 renders it slowly (though I think I know a way to significantly accelerate the rendering of these massive models.)

--Chris

Chris,

I may be interested in the "big one", as a test. But if you can find a way to optimize it, it could also be a good way to prepare for the "next things to come" !

By the way, with your released models of the Itokawa, can we apply a texture on the mesh ? I want to use some smooth texture so I don't get a boring gray piece of rock in space.

Cham wrote:Chris,

I may be interested in the "big one", as a test. But if you can find a way to optimize it, it could also be a good way to prepare for the "next things to come" !

By the way, with your released models of the Itokawa, can we apply a texture on the mesh ? I want to use some smooth texture so I don't get a boring gray piece of rock in space.

There are no texture coordinates on the mesh . . . I could probably modify my txt2cmod tool (now checked into CVS) to generate them with a spherical projection. However, there are likely to be problems along the meridian where the left and right sides of the texture meet.

--Chris

GEEZ ! The biggest one of your conversions is absolutly superb in Celestia. The FPS is pretty high, and the level of details is very impressive. If only it had a texture ! It's looking like a plaster "ball". It needs something...

Here's a link to the medium model (16 MB cmod file. 3.6 MB zipped !) :

http://nho.ohn.free.fr/celestia/Cham/Di ... 1.cmod.zip (3.6 MB zip file).

I need an SSC file for that dust pile.

EDIT : Don't download from this link. The model can be optimized. See below.

Chris, I converted the biggest model using the CMODTool to the text format (so I can play with the color) and then back to binary. The file size falled from 63 MB to 18 MB, without any apparent change in Celestia. What's going on ?

And for an interesting comparison, here's a picture of the real asteroid :


And here's the model in Celestia (the point of view isn't exactly the same) :


EDIT : WOW ! I've downloaded the 3 millions triangles version and tested it. The level of details is awesome ! I converted the model to text format, changed a bit the color, and back to binary cmod. The file size dramatically dropped from 252 MB to 72 MB ! And the FPS is very smooth on my system. This is just incredible. But what about the file size ? What's up !? Why was it so large, since I was able to reduce it to 72/252 = 29% of its original size !?

Here are two views of that 3 millions triangles puppy :

Cham wrote:By the way, with your released models of the Itokawa, can we apply a texture on the mesh ? I want to use some smooth texture so I don't get a boring gray piece of rock in space.

You could always paint each face with it's own individual color in your modeller. See you back here in a few months.

Chuft-Captain wrote:

Cham wrote:By the way, with your released models of the Itokawa, can we apply a texture on the mesh ? I want to use some smooth texture so I don't get a boring gray piece of rock in space.

You could always paint each face with it's own individual color in your modeller. See you back here in a few months.

That idea isn't even that far off!
If there is no real texture available, Celestia could assign a random color (based on a user specified color + some parameters perhaps) to each vertex of the mesh... That should be enough for asteroids?

steffens

Get the (x,y,z) coordinates of each vertex, pass them to a Perlin noise function, map the output of the noise function to some kind of colour gradient, and you're done.

Some C code for Perlin functions, and a general description of what they are, here.

Cham wrote:Chris, I converted the biggest model using the CMODTool to the text format (so I can play with the color) and then back to binary. The file size falled from 63 MB to 18 MB, without any apparent change in Celestia. What's going on ?

I forgot to specify the --uniquify option for cmodfix, resulting in lots of extra vertices in my models. I've reprocessed my meshes and zipped them--compressed, the high detail model is now just ~43M.

--Chris

EDIT : Here's my SSC definition for Itokawa :



Please, feel free to propose any modification or corrections to this SSC code. If you find more reliable data for the Itokawa orbit, tell us here. In particular, I'm not sure of the size (I've found many dimensions for this asteroid on the net).

Sorry to dig up old projects, but I'm just curious to know if anyone
ever derived a texture to go with this model?

This is one weird puppy. I've been playing with it for a day or two now,
and it never seems to behave as any other asteroid model does.

Why is this? Just curious (as always.)

Thanks, Brain-Dead

BobHegwood wrote:Sorry to dig up old projects, but I'm just curious to know if anyone
ever derived a texture to go with this model?

As far as I know, there's no albedo map for Itokawa from which to create a texture. Too bad, because some images (natural color, I think) from Hayabusa show dramatic color/brightness variations across the surface.

This is one weird puppy. I've been playing with it for a day or two now,
and it never seems to behave as any other asteroid model does.

Why is this? Just curious (as always.)

What do you mean by 'behave'? It does look different than other asteroids because it's by far the smallest asteroid we have good images of, and I think that this difference in scale makes it's 'rubble pile' nature more evident than in larger asteroids.

--Chris