Good Mesh needed for Phoebe

[t00fri]

Hi,

I am about to commit a neat Phoebe texture that I have extracted for
Celestia from the recent simple cylindrical Cyclops maps of
various Saturnian moons. For now, we have to use
'roughsphere.cms' as a generic mesh for Phoebe.

+++++++++++++++++++++++
But how about our model experts?
Wouldn't that be a worthwhile challenge??
+++++++++++++++++++++++

You may start from here:



I also did some significant update of Phoebe's orbiting
parameters that now allows an ~ quantitative reproduction
of the various photographic views of Phoebe from Cassini!
Partly, I used data that Grant found in recent scientific
publications.

Moreover, I was playing for quite a while trying to
reproduce the various Phoebe views from the Cassini flyby between
June 7 and June 11, 2004 for the quoted distances in

http://saturn.jpl.nasa.gov/multimedia/images/image-details.cfm?imageID=911
http://saturn.jpl.nasa.gov/multimedia/images/images.cfm?startImage=17&categoryID=4&subCategoryID=13

Finally, I succeeded to reproduce all four of the
characteristic views for distances between 12500
km (June 11) and 2 500 000 Km (June 7) with the
following data:

Code: Select all

RotationPeriod          9.2735 # Bauer et al, Astrophysical Journal 2004; 610(2): L57-L60   
Obliquity               8.7    # Porco et al, Science 2005; 307: 1237-42   
EquatorAscendingNode  245.3    # Porco et al, Science 2005; 307: 1237-42   
RotationOffset        135.0    # approximate, from encounter images


The first three are from Grant (i.e. from the respective
papers). The important modification I need is in the
RotationOffset: 135 degrees! The tests are quite sensitive,
hence I am rather certain that my value is about right...

I showed some Phoebe images with my new texture
recently...

http://www.celestiaproject.net/forum/viewtopic ... ght=phoebe

For people who want to start modelling right away, here is
the 1k Phoebe texture that I'll commit to the 1.4.1
release



You may just download it from the image...Of course, I
have a 4k hires texture, too, but for the default distribution that
would be too big.


Bye Fridger

[buggs_moran]

I made a mesh. It seems to fit nicely over the topographic map you posted Fridger. However, I think you or I need to play with the positioning of the texturing some. Something doesn't seem right... Here is the mesh to play with http://www.clanmoran.net/temp/phoebe1.3ds Perhaps the mesh needs to be reoriented.

[t00fri]

I made a mesh. It seems to fit nicely over the topographic map you posted Fridger. However, I think you or I need to play with the positioning of the texturing some. Something doesn't seem right... Here is the mesh to play with http://www.clanmoran.net/temp/phoebe1.3ds Perhaps the mesh needs to be reoriented.

Many thanks buggs,

unfortunately there seems to be some inherent shift. Did you calculate with a RotationOffset of 135 degrees and the other recent parameters by Grant?

Bye Fridger

[t00fri]

Hi,

also Jestr sent a large (1.8Mb) cmod model that was done
with the most recent phoebe parameters. I can't resist
showing you a comparison image with that mesh and my
4k Phoebe texture from June 11, 2004 at 12500 km
distance from Cassini:




Notice the PERFECT coincidence of the view (corresponding
to my rotationOffset=135 degrees). There are still a few
details less than perfect (e.g. the large crater) but still is
MUCH better than roughsphere.cms.


Bye Fridger

[buggs_moran]

very cool, I bow to Jestr's extreme model...

[ElChristou]

I'm curious to see the technic used by both (Buggs, Jestr); can you post 2 lines about this?
I'm also working on a model, with a preliminary stage around 120Ko...
The bases are some outlines extracted from the Nasa shoot provided by Fridger...

Edit:

here is a shoot in a similar presentation of the above doc; note that the lighning is too direct resulting in strong light/shadows; for a final 132ko model, I think it's not so bad...

[buggs_moran]

I am afraid my method was not too scientific, more artistic. I used the jpg in Fridger's post as a reference image, mapped an oblate sphere to its rough size and manipulated the mesh in each viewpoint until it matched the jpg. Then I smoothed it out a bit. I took the jpg literally as top being the north "pole" of Phoebe, which is why I dont think the texture mapped correctly. It's all good, maybe I'll use the 3ds model for another system at some point. It's available in the link from my above post as a 3ds.

[jestr]

Hi Chris,I just did it much the same as you and Buggs mate,working from the reference images Fridger posted above,also I think someone made a nice cylindrical map out of the 4 height scale views and posted it on the forum.I cant remember who it was made it though (maybe Steve Albers?).If you do a search you may find it.Cheers Jestr

[hank]

It seems like we should have been able to get a shape model for Phoebe from NASA by now, but no such luck I guess. The manual methods described here appear to give good results, but I thought I'd mention another idea for generating a shape model from the four-view image Fridger posted above, in case anyone wants to try it: note that each point on the surface of Phoebe appears in two of the four views. If the corresponding pixels could be correlated, then the three-dimensional coordinates of the point could de derived from the pair of two-dimensional image coordinates. This correlation could be done based on the pseudo-coloring of the pixels, taking advantage of scan-line coherence. Since the images were computer generated, the alignment is probably precise enough to allow this. It shouldn't be too difficult to write a small program to do it.

- Hank

[Toti]

The manual methods described here appear to give good results, but I thought I'd mention another idea for generating a shape model from the four-view image Fridger posted above, in case anyone wants to try it: note that each point on the surface of Phoebe appears in two of the four views. If the corresponding pixels could be correlated, then the three-dimensional coordinates of the point could de derived from the pair of two-dimensional image coordinates. This correlation could be done based on the pseudo-coloring of the pixels, taking advantage of scan-line coherence. Since the images were computer generated, the alignment is probably precise enough to allow this. It shouldn't be too difficult to write a small program to do it.

I did all this a long time ago using that same image as input and a small script I wrote. I don't have the model anymore (nor the needed tools), but in this thread:

http://www.celestiaproject.net/forum/viewtopic ... c&start=15

I posted detailed instructions and the script to
do almost all the job. It only takes a few minutes
and there's no manual modelling involved

The results are really good, apart from the rather
small distortion caused by the input image
limitations, as Grant and I agreed.
I corrected the problem with a single axis
stretching, IIRC.

The mesh file size was much less than 1.8Mb, btw.

[GlobeMaker]

Hello team,

My job at work is to make 3D meshes of planets and moons. One goal
for my product line is to produce many small moons of the outer planets
as plastic globes. A potential customer has asked me to do that for him.

Here is a method that may be of interest to you :
http://www.users.globalnet.co.uk/~arcus/mmps/mmps.html
That software takes the four views of a rounded moon and makes
a cylindrical projection array.

The new image can be put in a non-compressed format like .bmp format :
http://www.fortunecity.com/skyscraper/w ... ffrmt.html

From that .bmp array, using a custom program, the height array can be created according to color.
The height array then is used by my custom software to make a 3D
mesh.

However, the four color images from NASA have much less detail
than is needed for a high quality globe. That is why I am not pursuing
the Phoebe moon globe work. Even if NASA provided the height array
that created the 4 green moon views, there is much less detail than
seen in the texture pictures. The mesh would be very smooth while
the texture shows thousands of craters.

To get a high quality globe requires about a half million data points.
That size would show many crater shapes accurately. But that size
would have about 18 megabytes in .3ds format. Is an 18 megabyte
Phoebe model desireable?

[ElChristou]

I'm totally agree in having a much scientifiv approach; even if the work is good, an automated tool would be much better to avoid error of interpretation etc...

[t00fri]

Hi Chris,I just did it much the same as you and Buggs mate,working from the reference images Fridger posted above,also I think someone made a nice cylindrical map out of the 4 height scale views and posted it on the forum.I cant remember who it was made it though (maybe Steve Albers?).If you do a search you may find it.Cheers Jestr

Jestr,

no it wasn't Steve Albers, it was Matthew Arcus and his mmps software, and Toti, of course:

http://www.users.globalnet.co.uk/~arcus ... hoebe.html
http://www.celestiaproject.net/forum/viewtopic ... 6&start=35

There is also the result of Toti's procedural method to be seen.

Let me add that Grant has tried hard to get hold of the real
shape model. Peter Thomas dropped out of contact with
him shortly after agreeing to send it to Grant in November.


So we might as well have another try at PT

Bye Fridger

[t00fri]

Hello team,

...
Here is a method that may be of interest to you :

http://www.users.globalnet.co.uk/~arcus/mmps/mmps.html
That software takes the four views of a rounded moon and
makes a cylindrical projection array.
...

GM,

that method is very well known to me at least, notably
Matthew Arcus' mmps software, whith which I did a lot of
my work of generating cylindrical maps from Cassini
photographic imaging. Like my Titan, Iapetus and various
other textures now in the official distribution.


It was me actually who got Matthew interested in Celestia
and the typical problems we are facing here with texture
production


Bye Fridger

[t00fri]

Hi all,

Jestr is making AMAZING progress with his hires mesh for my Phoebe
texture that was derived from the recent Cyclops maps. I cannot
resist showing you the state of the art. Besides its beauty and strong
resemblance to the original (bottom), the Celestia image from the
Phoebe flyby is also a critical test for my revised orbital parameters.
Note how well Jestr has now modelled the big crater, and how precisely
the orientation fits at the nominal Cassini-Phoebe distance of 12500
on June 11 2004.



And here the corresponding Cassini Photo



Bye Fridger

[t00fri]

Hi all,

I quickly extracted the official USGS Phoebe locations (craters[AA] & regions[RE]) with a PERL script. Have a look below. There is only a problem with the crater "Phlias" that is correctly in my list, can be selected at the right place (marker with CTRL+K), but the label is not displayed. All the other labels are just fine.



Bye Fridger

[t00fri]

Hi,

Jestr just sent me another improved (cmod) version of his great Phoebe mesh. It's getting better every time Here is a neat composite showing that new mesh,my 4k phoebe texture, the feature names and the rising glorious Saturn!



Bye Fridger

[cartrite]

I'm not sure why, but the model I created always loades uoside down. Like This:



I created the model right side up. Does anyone else expierence this?

A couple of shots of it right side up.








cartrite

[jestr]

Hi Cartrite,here's what my mesh of Phoebe looks like in 3ds Max,with Grant's 'westgrid.jpg' applied as a texture.All moon or asteroid models have to be rotated 90 degrees so that their poles are pointing towards you when viewed from in front or behind in your modelling soft,and the 0 degrees longitude 0 degrees latitude point is facing you in the left window.



cheers,Jestr

[cartrite]

Thanks Jestr, Now this all makes sence. Before I figured out how to apply UV coordinates with Blender, I was using UV mapper classic, When I mapped in cylindrical the texture was not projected right. I was trying to apply a cylindrical map on a sphere with a polar Z. It should have had a polar Y. Thanks Again.

cartrite