Re: [Celestia-developers] Setting the size of objects in celx.

[vincent <vin...@fr...>]

Chris,

Actually, I had read the comment in stardb.cpp, but didn't precisely
understand the reason why a better management of the StarDetails
class was needed. Now, your explanation makes everything clear.
Thanks for taking the time :-)

I've checked in the change that adds the setradius method for a
solar system body.

As an example, the following script maps the [D] key to a function
that toggles the size of planets in whatever solar system:
http://vincent.gian.club.fr/celestia/planets_toggle_size.celx

Actually, this change already allows showing the cycle life of the
Sun dynamically. One just has to create a .ssc file to define a new
Sun, and make its radius change dynamically within a celx script.

@+
Vincent


Selon Chris Laurel <cl...@gm...>:

> Vincent,
>
> The change that allows setting the radius of a solar system object
> is good,
> but there are problems with the code to change star properties.
>
> In order to keep the size of star objects as small as possible,
> star
> properties are divided into two categories: properties that are
> part of the
> Star class, and properties contained in the StarDetails class. The
> Star
> properties are catalog number, position, and absolute magnitude,
> while
> everything else is in StarDetails: radius, temperature, shape,
> rotation
> rate, possibly an orbit. In general, stars with an identical
> spectral type
> will share the same StarDetails object. For the vast majority of
> stars, we
> don't have a lot of information beside a position, magnitude, and
> spectral
> type; they can thus be represented very compactly because all the
> StarDetails properties are shared with other stars of that spectral
> type.
> Yet customization is possible for those few stars where other
> information is
> known, e.g. the flattened shape of Altair, or the orbits of binary
> stars.
>
> Because the StarDetails object may be shared with other stars,
> naively
> adjusting the radius of a star is likely to change the radius of
> many stars.
> And unfortunately, there's no easy way of telling whether star's
> details are
> shared. Presently, everything is initialized at star creation time;
> if a
> star's radius, orbit, shape, or other extended info is specified in
> an .stc
> file, then the star will have it's own details object. There's a
> comment in
> stardb.cpp mentioning that better management of StarDetails objects
> is
> needed. Without it, dynamic adjustment of star properties isn't
> really
> possible.
>
> Additionally, changing the absolute magnitude of a star potentially
> requires
> rebuilding the star octree. A star that gets brighter is visible
> over a
> larger range, and may need to be moved into a larger octree node.
> Octree
> rebuilding is an expensive operation. It's not reasonable to do it
> every
> time the absolute magnitude of a star changes. Given these
> limitations of
> Celestia, I say you should defer your new functions for modifying
> stars
> until after 1.5.0, but go ahead and check in the function to set
> solar
> system object radii.
>
> --Chris