Phoebe and Wild2 new texture.When?

[danielj]

I would like to know if someone have an estimate about when a Phoebe texture will be released and when a Wild 2 new texture will be released.I think that the only thing that have to do with Wild 2 is to put that image in a map.I think that most people lost focus,because Celestia is not an Earth simulator,and for most of us,even higher resolutions of Earth will be set in very low frame rate.I would like what are the difficulties with textures for Phoebe and Wild 2?Does NASA have to process the image,or something like that?

[selden]

Daniel,

Do you want something accurate or something imaginary?

Both Phoebe and Wild-2 are very irregular. Someone has to create 3D models with the right shapes. Also, someone has to process the pictures to create appropriate surface maps. Often someone at NASA creates the models and someone at the USGS creates the maps. Both could take a very long time.

[granthutchison]

Daniel:
This sort of post is exactly the sort of thing that annoys hell out of many Celestia folk, myself included. What you have implied above is that those involved in texture and model generation are simply too lazy and unfocussed to get on with a trivial task - this is very rude indeed, even if you didn't make the post with that intention. It's exactly why people less polite than myself and Selden have been known to lose their temper with you in the past.
As Selden says, making a proper shape model (as opposed to a fantasy shape) is a tricky task, requiring some very precise information about spacecraft position and lighting angles which aren't usually generally available. Only then can a proper cylindrical map be generated to lay on to that shape model. And then we'll need to wait for such material to be published in a copyright-free form.

It's very evident than you're not prepared to take on this task yourself, so please don't speculate about the attitudes and motives of others.

Grant

[chris]

Someone on the Cassini team has created a model of Phoebe:

http://saturn.jpl.nasa.gov/cgi-bin/gs2. ... type=image

Unfortunately, the original geometry data doesn't seem to be available anywhere, and extracting it from the image, while possible, would require more time than I have. Does anyone know someone to contact about getting the Phoebe shape data?

--Chris

[jose21]

My guess is you'd contact someone at http://ciclops.lpl.arizona.edu/team/iss_team.php

They seem to be the people in charge of the image processing, so I guess they have the data.

[granthutchison]

Obviously a personal contact would be ideal. But meanwhile I notice that Carolyn Porco, the team leader, has an e-mail address at Ciclops that's obviously set up specifically for incoming comments arising from publicity releases. I've dropped a very brief request into it. Sometimes these things get a response, sometimes they are lost forever amid a thousand requests from schoolchildren doing projects. But it can't do any harm.

Grant

[granthutchison]

Sometimes these things get a response, sometimes they are lost forever amid a thousand requests from schoolchildren doing projects.

Well, she got back to me overnight, saying they don't want to release the raw derived shape data until they've published it as part of their initial report of scientific results from Cassini.

Grant

[chris]

Sometimes these things get a response, sometimes they are lost forever amid a thousand requests from schoolchildren doing projects.

Well, she got back to me overnight, saying they don't want to release the raw derived shape data until they've published it as part of their initial report of scientific results from Cassini.

I'm not too surprised, but thank you for taking the time to contact Carolyn Porco about this. It was good of her to respond promptly at least. In the mean time, I keep toying with ideas for reconstructing the model from the colors in that image . . .

--Chris

[Toti]

In the mean time, I keep toying with ideas for reconstructing the model from the colors in that image

Assuming that the shading of the source image can be eliminated:
Sometime ago I used a small free utility that takes a cube map and builds a cilindrical/spherical map with it. Then this map can be used to displace a sphere, and this mesh can be exported as a 3DS
We can think of the observer as being at the center of the body, looking at four square walls, ceiling and floor, where the body's surface is projected.
Thus:
1) We need 6 images (we only have 4: the ceiling and floor of the cube are missing. Should them be replaced with a midtone? Surely this would imply undesired effects in the final result)
2) The images must portray a 90 degree FOV shot. The source image doesn't fulfill this. Perhaps this can be fixed with some Gimp manipulation...

Another idea: are there enough shots of the Cassini flyby available? If so, after some image preprocessing, a contour of the body could be derived using a NURBS circle (or another similar tool). From a complete revolution, these "meridian" lines could be used to generate a NURBS surface, and therefore, via conversion, a 3DS. There are Open Source tools for this.

Bye

[jose21]

I'm sure you could process the images, but is it worth it? Meaning, if it's going to take a few weeks to do that, by then, the data might be available...

[Toti]

To clarify further my previous (rather cryptic) post:

Most 3D modelling programs allow you to place an image as background in the working window.

So:
1) Preprocess the set of images in order to make them of compatible size and centered (maybe using Cassini-Phoebe distance at the exposure time).
2) Threshold each image so it gets separated in black (space) and white (Phoebe).
3) In the 3D modelling program, create a circle of let's say, 50 vertices, with top and bottom vertices fitting Phoebe's contour.
4) Then, for each image:

1) Put it in the background of the 3D modelling program.
2) Make an in-place extrusion of the circle. Let's call this ecircle.
3) Move the ecircle's vertices horizontally (constrain to the window's X coordinate) to match the contour of the image.
4) Once done, rotate (appropiately!) the ecircle along the Z axis (so the model describes a true 3D object).
5) Rotate the viewport so it is now orthogonal to the ecircle's plane. (rotate it along the Z axis the same angle as in step 4)

5) Convert the obtained 3D surface mesh into a NURBS or Subsurfed one, and increase the resolution.
6) Reconvert it to a mesh, remove duplicated vertices and save it as .3DS

Bye

[Toti]

Well, I quickly tried my method using the four images posted by Chris, and I must say that the result is really interesting. Although that four images give only two useful contours, the general shape of the satellite is visible.
I used a 32 vertices circle, and the task took me about half an hour. The missing cross sections (at 45 and 135 degrees) were approximated to the default circle diameter (a poor compromise). It would be far better if I had two additional cross sections to work with

Bye