Page 1 of 1

artifact or hole?

Posted: 08.10.2003, 01:48
by steve.wray
Hi there!

I downloaded and installed a huge star database for celestia, its about 45M or so.

Something I noticed as I was 'flying around' the neighborhood of Sol was that there *appears* to be a big doughnut of stars around Sol with Sol pretty well slap bang in the center of it.

The torus is horizontal with respect to the plane of the galaxy and appears to have an internal radius of circa 500LY IIRC. The Hyades cluster is just outside this structure and over to one side.

Inside the hole in the torus the star density appears to be significantly lower than outside of it; ie Sol appears to be very isolated.

What I'm wondering is; is this an artifact of the star database or is this a genuine feature of our stellar neighborhood?

Thanks!

Posted: 08.10.2003, 02:30
by selden
Steve,

I don't seem to be seeing the effect you mention, but I might not be looking at things right.

The two large star databases that I have (downloaded from Pascal Hartmann's site at http://perso.wanadoo.fr/celestia.stars/index.html) are 17MB [sorry: 24.8MB; 17 is what XP shows it to be in the Zip file] and 50.6MB, not 45MB. Did you misremember the size or might your copy be damaged?

Assuming your copy of starsdb2.1C (the big one) is OK, can you provide a Cel:// URL for the viewpoint where you see this effect?

What limiting star magnitude are you using? 6.5? 15.5? something else?
Which star viewing format are you using? Scaled discs? Fuzzy points? Points?
Can you provide a screen snapshot of this effect?

It is true that there are some very bright clusters of stars some distance away, but there are lots of dimmer stars within 500LY.

Posted: 08.10.2003, 02:37
by steve.wray
The easiest quickest way I can suggest to get the view I saw, as I am at work...
is to go to Sol,
rotate the view so that you are over the top of Sol, perpendicular to the galactic plane,
and start zooming out.

If I'm right you should start seeing it once you are more than about 500LY out from Sol.

Got to have 'stars as points' to see it clearly.

I had maximum stars visible; magnitude and distance, though I turned distnce down to around 800LY to get best view.

When I get home I'll give more details

Posted: 08.10.2003, 03:03
by selden
Steve,

Given your instructions, I do see the effect in both the 24.8 and 50.6MB datasets. Very strange.

Here's a Cel: URL for the viewpoint:
cel://Follow/Sol/2003-10-08T02:01:54.61034?x=AAAAAIB3SFhPwpze/////w&y=CkxgAAA2oQDl9p8Q&z=VcpQAIDecr8ZYFUF&ow=0.703969&ox=0.059222&oy=0.652468&oz=0.274238&select=Sol&fov=43.293781&ts=1.000000<d=0&rf=66053&lm=0

Now it seems we need to get a professional astronomer to comment. Certainly there are bound to be some systematic errors due to the way distances are estimated from the Tycho catalogs, but still...

Posted: 08.10.2003, 03:44
by selden
and here's a snapshot showing the effect:

Image
(as usual, this links to a much larger image)

This is with StarsDB v2.1A (the 24MB version) and with "Fuzzy stars", looking from the viewpoint URL in my previous posting just above.

Posted: 08.10.2003, 04:23
by marc
If you look at Pascal's notes on how the tycho stars were extracted you can see this:
Number of Tycho stars with distance <= 25 pc : 0
Number of Tycho stars with 25 < distance < 50 pc : 44033
Number of Tycho stars with 50 <= distance < 100 pc : 43500
Number of Tycho stars with 100 <= distance < 500 pc : 520551
Number of Tycho stars with 500 <= distance < 1000 pc : 250659
Number of Tycho stars with 1000 <= distance < 2000 pc : 27177
Number of Tycho stars with 2000 pc <= distance : 1723


I think Pascal may have chosen this cutoff value to stop duplication closer than 80 ly from sol.
Or does the Tycho catalog not include any stars closer than 25 parsecs?.
I found this hole when looking for duplicated stars after i combined
Pascal's stars.dat with grants nearstars.
( http://www.celestiaproject.net/forum/viewtopic.php?t=2594 ).

Its not released yet but my MySQL addon can now output stars.dat and starnames.dat files, this should make hacking stars.dat files much easier.

Posted: 08.10.2003, 04:37
by selden
Marc,

A quibble: Pascal's selections are in terms of parsecs, while the dimension estimates of the hole are in light years. Since that's about a factor of 3.26 in length, I'm not sure they quite correspond.

Can you generate a histogram of star counts for volumes with radial boundaries of about 40LY? That might give us something more understandable to look at.

Thanks!

Written later: it'snot only that there's a hole, but also that there seems to be a significant excess outside it for a comparable distance. I think the extent of the excess is rather strange.

Posted: 08.10.2003, 05:54
by marc
internal radius of circa 500LY IIRC
Oops. Im thinking of another hole. I quickly assumed this was the same one.

Hopefully I'll be able to have a go at that histogram when i get home.

(added later), Sorry. I havn't found the time yet, I do want to investigate this and have put it on my todo list.

Posted: 09.10.2003, 03:51
by Paul
Closer to Sol, there are a great deal more faint stars picked up by the surveys.

My guess is that the "hole" is caused not by a lack of stars, but more by the star sprites being both denser and fainter (on average) closer to Sol. Being "blobs" rather than points, the sprites will render over each other - if more, fainter sprites are rendered over the brighter ones, the area will look fainter overall.

What could invalidate this theory is whether the star sprites are rendered using additive colour blending, in which case the fainter sprites wouldn't mask the brighter ones.

Cheers,
Paul

Posted: 09.10.2003, 03:54
by Guest
Even with 'stars as points'?

Posted: 09.10.2003, 20:44
by chris
What could invalidate this theory is whether the star sprites are rendered using additive colour blending, in which case the fainter sprites wouldn't mask the brighter ones.


Stars are indeed being rendered with additive blending, so the issue isn't fainter stars occluding brighter ones.

--Chris

Posted: 09.10.2003, 23:26
by Guest
Oh one other thing;
I get absolutely no trace of this phenomenon with the default stars.dat