Walls and Voids: >200K galaxies from 2dF (upd)

[selden]

More than 200,000 galaxies from the 2dF Galaxy Redshift Survey can now be seen in Celestia.

Both the complete dataset, and the 100K subset released in January of 2001, clearly show the walls and voids that have become so well known.

These Addons use the Celestia Model format and requires Celestia v1.3.2pre3 or later.

http://www.lns.cornell.edu/~seb/celestia/2df_100k-cmod.zip (1.8MB, updated 12:45 est 22feb04)

http://www.lns.cornell.edu/~seb/celestia/2df_300k.zip (4MB, 6Mar04)

There's no reason to install them both simultaneously, though. The 100K dataset is included in the 300K version.

For more information, see the Web page http://www.lns.cornell.edu/~seb/celestia/catalogs.html#3.5.7


(this thumbnail links to a much larger picture which shows the galaxy distribution in detail.)

[selden]

I found a bug in my position calculations for the 2dF model: the sign of the declination had been lost, placing all of the points in the northern hemisphere, when most of them actually are in the south.

Sorry.

The corrected zip file was uploaded at 12:49 today, 22 Feb.

While I was at it, I enhanced the model by color coding the northern and southern surveys. This revealed a problem in the original catalog data.
About 5200 galaxies have what seem to be spuriously small redshifts, and many of those have the wrong equatorial coordinates, too.

[Cham]

Selden,

can you make this addon workable in Celestia 1.3.1 ?

There's no 1.3.2 for OS X yet.

[selden]

Cham,

Unfortunately, so far I've been unsuccessful at
creating a 3DS model consisting only of glowing points.

I suppose I could create a multi-object version of the database similar to what I did with the QSO catalog, but trying to view 100,000 separate objects would really be slow.

[selden]

I've given up trying to make a 3DS file.

Here's the list of high-redshift galaxies as
Nebulas in a DSC file. They're all named "." so if you enable galaxy labels, you'll see their distribution in space. Since there are almost 100,000 of them, Celestia will be very slow.

http://www.lns.cornell.edu/~seb/celestia/2df_100k-dots.zip (1.6MB 22feb04)

The expanded DSC catalog file is slighly over 7MB. I've included an HTML file with a Cel:// URL so you can start Celestia looking at the dots from 7GLY away. The voids and walls are visible, but exploring them from different viewpoints will be, ummm, tedious.

[Cham]

Thanks Selden.

I tried the file. It works, but Celestia becomes SLOOOOOOOW !

This is sad, because this kind of data is VERY interesting.

How is it in Celestia 1.3.2 ? Is it smooth ?

When can I expect to see a Mac OS X version ?

[selden]

Cham,

It's impossible for me to know when 1.3.2 will be ready. It depends on several different projects getting finalized and enough bugs being stomped.

When a large database is used to construct a single 3D model instead of 100,000 separate ones, it's handled much more efficiently by Celestia and by the graphics hardware. It seems no slower than viewing the model of the ISS.

I'll try again this evening to generate a 3DS model. (12 hours from now)

[chris]

Selden,

What other line and point rendering features would you like to see in the cmod format? Support for depth cueing? Variable point and line sizes? Textured points (aka sprites)?

--Chris

[selden]

Chris,

You really shouldn't tease me like this

I can think of lots of features that would be useful, but I don't know how hard they are to implement, I hesitate to list them all, since it might scare you off

Some kind of depth cueing would be nice.

I'm rashly assuming this would consist of realtime luminosity changes due to changes in distance. One problem I anticipate would be determining how much is appropriate. (hmm. 1/r^2? scaled to some specified luminosity (in the model? in the catalog file? as a slider similar to the one provided for stars? scaled to 0 for the most distant element in the model? There's probably a better way, but I dunno what's feasable.)

Being able to specify the size of the point and line elements would be useful.

That'd help a lot with specifying the "importance" of different elements. Specifying the RGB values (as now) can be used to provide an intensity cue, but the more variables that are available, the more information can be provided in the display.

[added later: but how might this interact with depth cuing? Might it not imply changes of "diameter" with distance? -- perhaps only luminosity cuing initially?]

Sprites would certainly be useful, too.

I'm not sure what this implies, though. I don't know enough about OpenGL and 3D hardware to know how it would be implemented, but any of the possibilities I can think of would be helpful: An element with fixed pixel dimensions on the screen (thinking of ancient 2D graphics hardware)? Always face-on to the observer? Scaled and rotated at a fixed orientation relative to the underlying model?
Would something like this last concept diffeer a great deal from a texture mapped onto a set of vertices (onto a facet of the model)?

Give me some time to read over an OpenGL or 3D hardware primer, and I'm sure I can think of some more (I've gotta check on some local network problems right now.)

[Cham]

The only new features I'm really interested (for now) are these :

1-The ability to put stars anywhere, even in "a long time ago, in a far, far away galaxy".

2-BumpMap on any mesh.

3-Independant alternate surfaces for all objects. I mean, the ability to have an alternate surface for Io and another alternate object for Jupiter, for example, both at the same time.

4-Several light sources in the same solar system, for binary stars system, for example.

5-The ability to gives a custom texture to some stars, without adding an object around it.

6-and please, an updated Mac OS X version, without the preferences bugs ??

Sorry, this is off topic !

[selden]

Chris,

Oh, OK, after a Web search I think I found what you're thinking about: the OpenGL functions for point_size, line_size and point_sprite, modified by point_parameters.

Sure!


(Now if only I can figure out how to use such a sprite implementtion to do "real" particle systems)

[JackHiggins]

The only new features I'm really interested (for now) are these :

Animation!

Better rotation models for objects (3-axis stabilised spacecraft, in particular). Something like scripted control of what direction an object is pointing at, the ability to show attitude changes occurring, etc.. (like getting Cassini's HGA to track titan during radar mapping passes)

I've posted feature requests about this before! Would it increase interest in this feature if I said that ESA are particularly interested in this, to better represent their spacecraft being displayed at ESOC in Darmstadt?! This is something entirely realistic that celestia currently has no way to replicate (precession is not realistic).

Specmaps in the models would be nice as well.

[Kolano]

I'm digging up this thread to question if the visible galaxies flash in and out oddly for other people as the angle of view is changed/time proceeds. At times my display will go from showing many thousands of points to just a few hundred which is odd. It's irksome when one is trying to view things from a certain angle and most of the display disapears.

[selden]

Kolano,

It would help if you specified your system and software specs. See the "Preliminary User's FAQ" in the Users forum for an example.

Written later:

To put it another way, I haven't seen that probem. There are (were?) some clipping problems for some model shapes, but I've been careful to include invisible points at the boundaries of the distribution to try to avoid that problem and to provide the right scaling in Celestia.

Also, there were some symptoms similar to what you seem to be describing when displaying the catalogs of globular clusters, galaxies and galaxy clusters in older versions of Celestia, but I'm pretty sure they were fixed in v1.3.1. I'm not seeing them in the current prerelease.

System:
256MB 500MHz P3, Win XP Pro SP1
128MB GF4 Ti4200, Nvidia drivers v53.03
Celestia v1.3.2pre5

[Kolano]

OK. I'm running 1.3.2pre5 under Windows with a ATI 9800.

I have installed 2df_100k-cmod and quasars-cmod.

I would post a Cel URL to more explictly demonstrate the issue but that function seems to be broken in pre 5. I was able to partially rectify the situation by rememebering to de-sync from whatever I had been looking at before zooming out to veiw the galaxies, that at least stops the sparkling.

However, I still have an issue that as I zoom in and out. At some zoom levels 1000's of objects appear but at others only a about 100 do. There doesn't seem to be much rhyme or reason to it.

Could it just be that loading both files produces too many points for Celestia to handle?

[Cham]

You have an ATI 9800 ? How is Celestia with it ? Do you have any problems, like bump map, night textures, or anything else ?

I may buy one, in the next weeks, I'm not sure yet.

[selden]

Kolano,

I don't think that it's the number of points by itself that's causing problems. I've been working on some other databases that have many more points in their models, and I have not seen the problem you describe.

What version of the ATI graphics drivers are you using? Several bugs related to use with Celestia were fixed in their most recent update.

Have you tried changing the "render path" so that Celestia uses simpler functions to draw things?
Type a Ctrl-V several times. The first time, it'll switch to Basic, then Multitexture and then back to "OpenGL vertex program". Does viewing the database in Basic or Multitexture make any difference?

If it's not on your system any more, you might want to consider downloading the 1.3.2pre3 Celestia executable (not the whole kit; and give it a different name) just so you can use its "write url" function. It's available at http://216.231.48.101/celestia/files/celestia-win32-1.3.2pre3-exe.zip

[Kolano]

Cham:
I don't have any problems under Celestia except the one I've been describing with the latest release, so I'm fairly satisfied with my 9800. The one thing that's been annoying me is that for some reason TV-out support seems to disapear (the options for it are completely removed from the control panels). Sometimes a reboot fixes it, but often it requires a driver reinstall which is stinks.

Seldon:
I'm running ATIs latest drivers 4.2

Different render paths have no effect.

Maybe I'll try to generate some Cel URLs tommorow if no one can think of anything from what I've described so far.

[selden]

The "2dF Galactic Redshift Survey" final data release (Colless et al, 2003) is now available in a format compatible with Celestia v1.3.2pre3 or later.

You don't need both the 100K and 300K versions.

http://www.lns.cornell.edu/~seb/celestia/2df_300k.zip (4MB, 6Mar04)

See the top of this thread for more details.

[don]

Hi Selden,

FYI, just tried this one on my ATI card and it only works if I reduce the radius from E+10 to E+7.

Looks NICE!

Good work!