Celestia Forums

Hi,

Why do all selected asteroid orbit seem not to be drawn on gravity center of the objet? It is the case for instance of Ceres...

Thanks in advance.

Hi, Imy. For convenience, Celestia draws orbits not as ellipses but as polygons inscribed in the orbit. As a consequence, if you look closely at a body, you will see it wobbling closer and farther to its drawn orbit. Small bodies, such as asteroids, have their orbits outside of their bodies most of the time.

By default, Celestia draus orbits as 100-gons. You can change this setting by editing celestia.cfg. the variable is OrbitPathSamplePoints. there are other rendering variables as well, such as RingSystemSections, since rings are also rendered as 100-gons by default.

The true orbits, that is, the paths followed by the celestial bodies, are nevertheless elliptical.

Guillermo

Out of interest, how are the orbit points distributed - surely for best results the distribution should be based on the curvature of the ellipse.

ajtribick wrote:Out of interest, how are the orbit points distributed - surely for best results the distribution should be based on the curvature of the ellipse.

Yes, as of version 1.5.0, ellipses are adaptively sampled. You can see dramatic improvement with comet orbits. Also, 1.5.0 performs cubic interpolation between sample points. You'll find that small bodies lie much closer to the rendered trajectory than in 1.4.1. There's still room for improvement--I want to implement a better adaptive spline rendering scheme soon.

--Chris

And, as Guillermo has already pointed out above, you CAN create
smoother (and more accurate) orbits by modifying your celestia.cfg
file.

Just open it with Notepad and have a look at the documentation there.

Brain-Dead

chris wrote:

ajtribick wrote:Out of interest, how are the orbit points distributed - surely for best results the distribution should be based on the curvature of the ellipse.

Yes, as of version 1.5.0, ellipses are adaptively sampled. You can see dramatic improvement with comet orbits. Also, 1.5.0 performs cubic interpolation between sample points. You'll find that small bodies lie much closer to the rendered trajectory than in 1.4.1. There's still room for improvement--I want to implement a better adaptive spline rendering scheme soon.

--Chris

Chris wrote:- Cubic splines are used to reduce the appearance of sharp angles in between
orbit segments. This also has the effect of placing the rendered orbit path
much closer to the the actual orbit path.

Aha, good...

http://www.celestiaproject.net/forum/viewtopic ... line+orbit

F.

Imy wrote:Hi,

Why do all selected asteroid orbit seem not to be drawn on gravity center of the objet? It is the case for instance of Ceres...

Thanks in advance.

image ceres HUBBLE

http://hubblesite.org/newscenter/archiv ... iew:images

t00fri wrote:

chris wrote:

ajtribick wrote:Out of interest, how are the orbit points distributed - surely for best results the distribution should be based on the curvature of the ellipse.

Yes, as of version 1.5.0, ellipses are adaptively sampled. You can see dramatic improvement with comet orbits. Also, 1.5.0 performs cubic interpolation between sample points. You'll find that small bodies lie much closer to the rendered trajectory than in 1.4.1. There's still room for improvement--I want to implement a better adaptive spline rendering scheme soon.

--Chris

Chris wrote:- Cubic splines are used to reduce the appearance of sharp angles in between
orbit segments. This also has the effect of placing the rendered orbit path
much closer to the the actual orbit path.

Aha, good...

http://www.celestiaproject.net/forum/viewtopic ... line+orbit

F.

There were a host of new improvements to orbit rendering in 1.5.0. It's an incredibly complicated task to show smooth orbits that correctly occlude/are occluded by planets over a huge range of scales. A large part of the complexity is the limited arithmetic and depth buffer precision of graphics hardware.

- The scene is split into different depth intervals, roughly logarithmically. Each depth interval has a far plane that is not more than 1000 times the distance of the near plane. This is necessary to avoid depth buffer precision problems. An orbit may be rendered multiple times per frame, once for each span it covers.
- Orbit paths are cached in order to avoid the potentially huge overhead of resampling every frame. The cached path is split into sections (groups of segments) and per-section bounding volumes are computed fast culling of parts that lie outside the depth interval or the field of view.
- At render time, the orbit is transformed into camera space and clipped against the near and far plane on the CPU. Double precision arithmetic is essential for proper rendering, which is why transformation can't be done on the GPU. Orbit path segments are further subdivided using spline interpolation whenever they lie sufficiently close to the observer.

The culling of orbit paths by section is big part of why spline interpolation is practical now. That, and the fact that all the transformation has to be done on the CPU anyway because of the double precision requirement.

--Chris

symaski62 wrote:

Imy wrote:Hi,

Why do all selected asteroid orbit seem not to be drawn on gravity center of the objet? It is the case for instance of Ceres...

Thanks in advance.

image ceres HUBBLE

http://hubblesite.org/newscenter/archiv ... iew:images

http://dawn.jpl.nasa.gov/

arrival -1310 DAY VESTA

^^

I don't precisely understand how orbits are built : you use 3 sides polygons (like Kepler...) where one vertex is on the barycenter and two others draws a piece of the orbits... Then if you multiply the number of rod, it seems that you're orbit is a real curve, isn't it? One tip was to use spline instead of ligne so that the triangle become like an icecream basic shape? Where accuracy is lost? why double precision?
It is a solution to increase basic config number of polygons but i think this was selected to this value because above you need a good computer...

But why not starting on the gravity center of the planet and then continuing from there to close the orbit. Then it would be exact (from the object), no?

Is Celestia able to draw no-closed orbits like parabolic trajectories of comets?