Astronomy Picture of the Day recently featured a nice starfield image made
available at 6000x3000 by The European Southern Observatory (ESO) under a
Creative Commons license.  It appears to my untrained eye work very well as a
spherical image_map applied to a sky_sphere.  If any of you celestial mechanics
would like to point out how to calculate dw_startrot properly (as is done in
sunpos.inc) I'd appreciate it.

#declare dw_latitude=29.53;  // Observer's latitude.
#declare dw_startrot=65;     // Found by "hand calibrating" to a star chart
                             // for 2009-09-12T07:16 CDT at this location.

sky_sphere {
   pigment {
      image_map{
         png "phot-32a-09-fullres.png" // High resolution starfield image.
         //png "constellations.png"    // Image with constellations labeled.
         map_type 1                    // 0=planar, 1=spherical,
                                       // 2=cylindrical, 5=torus
         interpolate 4                 // 0=none, 1=linear, 2=bilinear,
                                       // 4=normalized distance
         once                          // No repetitive tiling.
         }
      scale  <-1,1,1>                  // Reverse projection handedness.
      rotate <118.5,0,-152.8>          // Put Polaris at +y.
      rotate  y*(dw_startrot+360*clock)// Rotate around polaris.
      rotate  x*(90-dw_latitude)       // Adjust for latitude.
      }
   }

Interpolation 0 seems best for most situations, though 4 is good for some.  No
interpolation makes stars behind a partially transparent layer either present or
absent, and seems better for dark skies, while 4 causes them to fade nicely when
behind another partially transparent pigment and seems better for lit skies.
Interpolation 1 does not seem to work, and 2 is in-between 0 1nd 4 but may make
it easier to pick out constellations in some situations.

The ESO's big image I converted to PNG since their TIFF didn't seem to work:
http://www.eso.org/gallery/d/133707-4/phot-32a-09-fullres.tif

I got the constellation-marked version by clicking "See the Constellations!" and
downloading the resulting image from:
http://www.gigagalaxyzoom.org/B.html

For information about "The Milky Way panorama" see:
http://www.eso.org/gallery/v/ESOPIA/Galaxies/phot-32a-09-fullres.tif.html

The Astronomy Picture of the Day that led me to it:
http://apod.nasa.gov/apod/ap090926.html

Serge Brunier did the original, and will reportedly send a very high resolution
version on request:
http://www.sergebrunier.com/gallerie/pleinciel/index-eng.html

Image map code shamelessly ripped from:
http://news.povray.org/web.4a5e5e8141a4106c6a9471340%40news.povray.org

Post a reply to this message

> If any of you celestial mechanics would like to point
> out how to calculate dw_startrot properly (as is done in
> sunpos.inc) I'd appreciate it.

I'm not a celestial mechanic but gave it a shot ;) The original
image seems to use the galactic coordinate system so I tried to
go from there. The source file is posted in p.t.s-f:

http://news.povray.org/povray.text.scene-files/thread/%3C4ac0e0cb%241%40news.povray.org%3E/

It's quite a bit longer than your version, though ...

Post a reply to this message

Christian Froeschlin wrote:

> It's quite a bit longer than your version, though ...

Thanks!  That looks like just the thing to do it.  :)

Unless you or somebody else beats me to it, I'll see if I can package it all up
for inclusion in the object library when I get more than a few minutes at a time
to rub together (if that's alright).

P.S.  I think I'll leave the galaxy at its default settings, at least for now.
;)

Post a reply to this message

waggy wrote:

> Unless you or somebody else beats me to it, I'll see if I can package it all up
> for inclusion in the object library when I get more than a few minutes at a time
> to rub together (if that's alright).

No problem, if you think it's generally useful ... when you
package it up, note that the HourAngle macro still contains
a lot of unnecessary code as a result from copy and paste.

You can remove the line

   #local e= 0.016709-1.151E-9*d;

as well as the entire code block from

   // Obliquity of the ecliptic

up to

   #local RA = degrees ...

as that is only relevant for the sun, not distant stars.

Did you test it against your original star chart? The only
test I did was to see if the vernal point would rise due east
at spring equinox 6am so I hoped it was correct. But I did a
lot of trial and error with the signs: mixing three celestial
coordinate systems with POV-Ray's left-handed one caused a
bit of brain overload at times ;)

Post a reply to this message

Christian Froeschlin wrote:
[...]
> Did you test it against your original star chart? The only
> test I did was to see if the vernal point would rise due east
> at spring equinox 6am so I hoped it was correct. But I did a
> lot of trial and error with the signs: mixing three celestial
> coordinate systems with POV-Ray's left-handed one caused a
> bit of brain overload at times ;)

Thanks for the cleanup tips.  I've only had time to do a quick test, but only
one small change in the rotation calculations was needed for everything to line
up a sky_sphere properly for my test date and time.

Toward the bottom, the line...

  #declare A_LOCAL = HourAngle(YEAR,MONTH,DAY,HOUR,MIN,LONG,LAT,LONG);

....is now...

  #declare A_LOCAL = HourAngle(YEAR,MONTH,DAY,HOUR,MIN,LSTM,LAT,LONG);

....with LSTM set as Lstm is in sunpos.

I'll finish recasting your code as a macro and post it as time permits.

Also, although I'm using this for a current fairly small animation, my primary
interest is for a long-term POV-Ray project that could be described as a
5,000-year time-lapse sequence, so I'll be looking into precession corrections
later.

Again, I thank you for this stellar (pun intended) example of the wonderful
support within the POV-Ray community.

Post a reply to this message