Page 1 of 1

Re: precession

Posted: 30.01.2002, 10:05
by dab
[ this is in reply to a post on the old board ]

dramsey wrote:
> Ayup, it doesn't do precession. For the simple reason that continually recomputing the position of 100.000 * stars for every redisplay we do (25 a second if you're > > getting a good screen rate), is NOT something you want your computer doing.

I do not think this is correct. After all, the program is not calculating a geocentric model
of the solar system. (which would be an interesting, if academic, exercise: think epicycles)
Of course, the "position" of the stars, at least the ones close enough
to be visible has to be recomputed any time you press a cursor key, but this is something
I very much do want my computer to be doing. The accounting for precession would be one
simple operation, namely aligning the earth axis with a point on a circle. And this only if
earth is close enough to be rendered as more than a dot or if "sync orbit earth" is on.

>Also, over that timeframe, stars also move acoss the starfield, and computing those positions is NON-trivial to impossible as the motions aren't really known very
>well. So, Celestia just doesn#t even attempt to do it.

this on the other hand is in agreement with what I have supposed: I have combined the two
issues in one post because I was thinking about long-(millennia)-time issues. Here, I very
much agree it cannot be done because the vectors are not known. I don't think it would
matter very much performance-wise, since the positions of the stars would only need to be
updated very sporadically.


cheers,

DAB

Posted: 30.01.2002, 18:40
by chris
Quite right . . . precession is something that we could model without too much trouble, but currently we pretend the Earth's rotation axis is constant.

Modeling the proper motion of the stars is more difficult. The HIPPARCOS database does include proper motions, but the motions aren't given in three dimensions. Even if they were, there's the additional overhead of storing them (an additional 1.2 megabytes for the current database) and updating the star positions. Celestia uses an octree for fast visibility testing of stars; if the stars moved, this structure would have to be constantly modified and reoptimized for cache coherence in order to maintain the same level of performance. It's probably not impossible though.

--Chris