Celestia Forums

Robert Gaskell's shape model of Vesta is now available. As with Gaskell's other shape models, this one is derived from spacecraft images (in this case, from Dawn.) I'm almost done with an add-on for Celestia. Here's a video comparing the shapes, sizes, and rotations of Phoebe, Vesta, and Mimas:

http://www.youtube.com/watch?v=dSGh_JHNFSI

--Chris

Chris,

have you repaired the seams problems on the other models ? And what about Itokawa with its holes ?

WHERE is it available?
It isn?t in Celestia Motherlode or in the index of extras...

The Vesta add-on is here:
http://www.celestiaproject.net/~claurel/celest ... /vesta.zip

I've also updated the Gaskell models add-on. It now includes Vesta, Eros, Itokawa, Phoebe, Phobos, and Mimas. If you download this one, you do not need the separate Vesta add-on above. The models are available here:
http://www.celestiaproject.net/~claurel/celest ... models.zip

Cham: I was able to reduce the 'seam' effect on Vesta somewhat, but it hasn't been completely eliminated. I'm not sure what to do except to try a different normal map generation tool, and I don't know of an alternative to xNormal.

--Chris

chris wrote:The Vesta add-on is here:
http://www.celestiaproject.net/~claurel/celest ... /vesta.zip

I've also updated the Gaskell models add-on. It now includes Vesta, Eros, Itokawa, Phoebe, Phobos, and Mimas. If you download this one, you do not need the separate Vesta add-on above. The models are available here:
http://www.celestiaproject.net/~claurel/celest ... models.zip

Cham: I was able to reduce the 'seam' effect on Vesta somewhat, but it hasn't been completely eliminated. I'm not sure what to do except to try a different normal map generation tool, and I don't know of an alternative to xNormal.

--Chris

The 2nd link seems to be broken. I get a 404 error message.

Reiko wrote:The 2nd link seems to be broken. I get a 404 error message.

Sorry about that... It's fixed now.

--Chris

I tried the model and texture and looks like something is wrong with Vesta.Its texture is bland
http://imageshack.us/photo/my-images/81 ... ctlyt.jpg/

danielj wrote:I tried the model and texture and looks like something is wrong with Vesta.Its texture is bland
http://imageshack.us/photo/my-images/81 ... ctlyt.jpg/

That's how the model is supposed to look: there is no texture applied other than the normal map. It's possible to apply a texture, but there's no global map of Vesta containing exclusively albedo variations. Take a look at this global map of Vesta from Dawn images:

http://photojournal.jpl.nasa.gov/jpeg/PIA14703.jpg

You can see that most of the brightness variation is caused by lighting and shadows. Such a map is a bad choice to use with high resolution geometry and/or a normal map: the static lighting in the texture map interferes with the dynamic lighting calculated by Celestia and OpenGL. What we really want is a texture with only that brightness variation caused by differences in surface materials. Unfortunately, such a map isn't available, and it's a challenging task to create one.

--Chris

But the other problem is that the textures files are in dxtnm format.Celestia only recognizes jpg,png and dds formats.Am I wrong?

danielj wrote:But the other problem is that the textures files are in dxtnm format.Celestia only recognizes jpg,png and dds formats.Am I wrong?

Yes you are wrong. .dxt5nm is a dedicated high-quality compressed normalmap format that Celestia supports. There are only normal maps in the textures/medres folder of ChrisL's add-on. No textures.

Fridger

But there is a Vesta texture.I downloaded it from unmanned spaceflight.It?s not a perfect texture,but it is a texture anyway.Is it wrong?When I got home,I will post a image of Vesta with this texture.But it will be only at night (after midnight,in Europe)

Here it is
http://imageshack.us/photo/my-images/85 ... mdawn.jpg/
Maybe the model is not correct...

danielj wrote:Here it is
http://imageshack.us/photo/my-images/85 ... mdawn.jpg/
Maybe the model is not correct...

I didn't claim that there is no Vesta texture in this world. I just wrote that there is NONE in the textures/medres folder of ChrisL's add-on.

Incidentally, your texture doesn't look too bad?

Fridger

danielj wrote:Here it is
http://imageshack.us/photo/my-images/85 ... mdawn.jpg/
Maybe the model is not correct...

It looks like you are using this map:

http://photojournal.jpl.nasa.gov/catalog/PIA14703

But, I think that you are using it incorrectly. Because Vesta's northern hemisphere has been in shadow for the duration of Dawn's visit, the map only has the equatorial and southern latitudes. If you just use the map directly, features will not be positioned correctly on the model. Notice that the 'Snowman' crater triplet should be near the equator, but it appears far to the north in your image.

The remarks I made earlier about the need for an albedo-only also still apply.

--Chris

Daniel,

the texture of size 2137 x 780 from
http://photojournal.jpl.nasa.gov/catalog/PIA14703
does not have the correct 2:1 aspect ratio, which corresponds to the fact that it only extends in latitude from -90 degs to about +41.5 degrees (rather than +90 degrees). ChrisL pointed out the reason for this. So, if you want to experiment with this texture, you simply have to add a uniformly black or grey texture strip of size 2137 x 288 pixels on top. This then complements the texture to a required aspect ratio of 2:1.

The central meridian is already correct, whence you don't have to rotate the texture horizontally.


Fridger

EDIT:
There is actually an erroneous, extra black 1 pixel column at the right hand end of the original texture that needs to be cropped, such that the correct size of the raw texture is 2136x780. You then add a black strip of size 2136x288 at the top, to arrive at a total size of 2136x1068, i.e. an aspect ratio of 2:1 as needed. You may also want to erase the texture owner's name in the lower left and fix the rounded black edges at the bottom.

I am confident that you will be able to perform these straightforward manipulations...

EDIT2:
I quickly performed the described manipulations and tryed the texture out. It seems that ChrisL's model requires a cube-mapped texture rather than a simple cylindrical projection. I am sure he can clarify this.

With the Celestia 1.6.1 default model, the texture looks nice, however. See here:
[Click on image by all means!]

Just to say this image is used for an add on made by ParticleGrasp as an Alt surface. It is on the Motherload site.
David.

t00fri wrote:EDIT2:
I quickly performed the described manipulations and tryed the texture out. It seems that ChrisL's model requires a cube-mapped texture rather than a simple cylindrical projection. I am sure he can clarify this.

The normal map was generated from the high resolution model using the free tool xNormal. I was free to use whatever mapping I liked for the generated map. I chose cube faces instead of the usual cylindrical mapping for two reasons:

1. To avoid the 'polar pinch' artifacts when rendering
2. To avoid any possible problems with generation of the normal map near the poles. I didn't actually test this, but I'm not optimistic that xNormal will do a good job handling the extreme distortion at the pole.

--Chris

Thanks Chris,

for confirming my guesses about the required cube-maps, if one wants to add a surface textures to your models. Given a good vesta model, I am still undecided between these two alternatives:

1) hires details via xNormal but NO surface texture and
2) using a hires surface texture without xNormal (eliminating polar pinches as I demonstrated for
Phobos at CM).

In my view the latter choice (2) still gives by far the more natural views, while the former one provides by far the best 3d details including varying illumination. It seems quite improbable to me, however, that soon flat cube-mapped surface textures will be availble for these non-spherical bodies...

The (necessarily uniform) coloration of your shapemodels looks quite unnatural. I did some detailed comparisons with recent (almost) natural color images of Vesta shown at the Geological Society of America meeting, Oct 2011, e.g.

As usual, the hires texture I prepared today, was computer-matched in color to this image. The screenshot below was based on the default 1.6.1 shape model of Vesta. 3d effects would look much better with your Gaskell model (made to accept cylindrical textures). The old derfault model is just far too round/smooth almost everywhere.
[Click on image by all means]


Fridger

t00fri wrote:Thanks Chris,

for confirming my guesses about the required cube-maps, if one wants to add a surface textures to your models. Given a good vesta model, I am still undecided between these two alternatives:

1) hires details via xNormal but NO surface texture and
2) using a hires surface texture without xNormal (eliminating polar pinches as I demonstrated for
Phobos at CM).

I'll try generating a version of the Vesta model that uses a cylindrical texture mapping. It shouldn't be too difficult. Hopefully, the cylindrical map is well-aligned with the 3D model (I mention it, because I recall reading somewhere that the map uses a new planetographic coordinate system.)

--Chris

chris wrote:

t00fri wrote:Thanks Chris,

for confirming my guesses about the required cube-maps, if one wants to add a surface textures to your models. Given a good vesta model, I am still undecided between these two alternatives:

1) hires details via xNormal but NO surface texture and
2) using a hires surface texture without xNormal (eliminating polar pinches as I demonstrated for
Phobos at CM).

I'll try generating a version of the Vesta model that uses a cylindrical texture mapping. It shouldn't be too difficult. Hopefully, the cylindrical map is well-aligned with the 3D model (I mention it, because I recall reading somewhere that the map uses a new planetographic coordinate system.)

--Chris

Thanks! That should be very useful for gaining more experience with this issue.

Fridger