M104 - Sombrero Galaxy

[jim]

Hi all,

I've made for Selden's billboard a 2k texture of the Sombrero Galaxy. But I have absolute no idea how to find the correct values for "Axis" and "Angel". The rest should be correct because is simple copied from Selden's 'messier_galaxies.dsc'.

Code: Select all

# M104 / NGC 4594, The Sombrero Galaxy
Nebula "M104"
{
   Mesh "m104.3ds"
   Axis [-0.20726 0.975903 -0.0682352]
   Angle 229.072
   RA 12.665
   Dec -11.62
   Distance 6.52e7
   Radius 70000
   InfoURL "http://simbad.u-strasbg.fr/sim-id.pl?Ident=M+104"
}


Selden, I think you are the expert. Can you help me? Here is the addon only with a lowres texture at the moment.

http://www.celestiaproject.net/~jim/files/m104-test.zip (158kb)

Selden, there is still another problem. I'd installed your Horsehead and Pr?spe's IC434 addon without to know that this is the same. Now the problem is that they are different orientated and I have absolute no idea which is correct. Can you check this?

Bye Jens

[selden]

Jens,

If you haven't already, please take a look at http://www.lns.cornell.edu/~seb/celestia/billboard.html#6.0
It describes how to use Celestia's internal orientation controls.

The best way to examine a model's position is to compare what Celestia shows from a viewpoint near the Earth with how it looks in an astronomical photograph.

If the model's surface texture includes the right stars, you may be able to compare the locations of those stars with the Hipparcos stars in Celestia.

If you look at my "billboard" of B33 (Barnard 33 = Horsehead Nebula; IC434 = large pink emission nebula behind it), you'll find that its stars match Celestia's stars quite well.

Unfortunately, getting the stars to align properly is extremely tedious with the commands currently available.. (Run celestia, edit DSC file, restart Celestia, edit DSC file, restart Celestia, etc.) It'd be a big help to Nebula designers if Celestia could be improved with Mouse commands to edit objects' RA, DEC and Radius the way the rotations are edited now. (hint, hint )

I suspect that Praesepe simply didn't have the time to perfect his Addon's orientation.

Added later: Sorry, I didn't mention that I'll take a look at your Addon this evening after about 6PM EST (= 23:00 GMT).

Also, Praesepe's DSC files include no orientation commands, which means that the alignment was done by changing the model's orientation within the 3D model editor. Matching those angles to Celestia's must have been at least as tedious as changing the numbers in Celestia: tweak model, export model, restart Celestia, tweak, export, restart, tweak...

[jim]

Hi Selden, thanks for the advice. But I have sill a problem I can't make Celestia to save the values. I start Celestia from commandline (the 'align.dat' is written) go to Edit mode '@' and can rotate the selected object. Now I press 'shift-!' or 'shift-1' = '!' but nothing happens. Which character exact does Celestia expect to write the values to the file?

Bye Jens

[selden]

Jens,

Shift-1 = ! on a U.S. keyboard is what writes the values to the log file.

The values are not written to the log file until Celestia exits.

[selden]

Jens,

Here are some parameters that seem to align your M104 addon:

Code: Select all

Nebula "M104"
{
   Mesh "m104.3ds"
Axis    [-0.57969 0.613711 0.536021] Angle 142.397

   RA 12.66645
   Dec -11.6225
   Distance 6.52e7
   Radius 91200
   InfoURL "http://simbad.u-strasbg.fr/sim-id.pl?Ident=M+104"
}

There are no stars in Celestia, even in Pascal's 2M addon, that are close enough to align the model, so I had to use another picture to provide them. Here is a picture showing your model in Celestia in front of the alignment picture:

(this links to a much larger image)

The alignment picture was created by the SkyView server, using the "advanced" interface at http://skys4.gsfc.nasa.gov/cgi-bin/skvadvanced.pl

The three stars that are marked in blue were used to align the alignment pattern. The two circled stars were defined by me in an STC file to be at exactly the coordinates of the grid on the picture. The one marked with straight lines is in Celestia's Hipparcos catalog (HIP 61823). I moved and scaled the alignment picture until it matched with those stars. (Actually, I defined 4 stars to align the grid. The other two are out of view above the top of this picture.)

The three stars that are marked in green were used to align your model of M104 in front of the stars of the alignment picture.

Does this help?

[jim]

Hi all,

The addon is ready - thanks Selden.



You can find some interesting details...


There are two versions for download available:
4k DDS DXT5: http://www.celestiaproject.net/~jim/files/M104 ... 4k_dds.zip (8,1mb)
1k PNG: http://www.celestiaproject.net/~jim/files/M104 ... 1k_png.zip (907kb)

If your card support DDS textures but only 2k texture size (see OpenGL info) use my little DOS-script to downsize the 4k DDS file 'ngc4594.dds' to 2k or 1k.
http://www.celestiaproject.net/~jim/files/downsize_DDS.zip (49,8KB)

To install extract the zip-file to Celestias 'extra'-folder or to a sub folder.
If Seldens Messier-addon is already installed it may be usefull to delete or dissable (with #) the entries for M104 in 'messier_galaxies.dsc'.
Selden, why there are two entries for M104 in your 'messier_galaxies.dsc' ?

Have fun!

Jens

PS:

Hi nebula addon developer!

Did you have the same problems for this equation:

simple nebula picture = texture with alpha channel in front of a black background

The problem is that a texture with and without aplha channel (transparency) should look equal in Celestia.
I found a relative simple solution for this.
Open the simple nebula picture in PSD or Gimp. Make a copy in a second layer and disable this (is later needed to compare). Now change the picture mode to 'Lab colour' and make a copy of the Lab brightness channel (This will be the alpha channel). Than change the picture mode to 'CMYK colour' and fill the black channel with white. Now change the picture mode back to 'RGB'. Add a third layer as black background. Now use the copied Lab brightness channel as mask for the first layer (the modified picture). Add to the second layer (the copy of the original picture) a simple strip mask in that kind that the new and original picture is alternate seen. Now use gamma correction with the alpha channel of the first layer and adjust this until no differnce can be seen between both pictures. Thats all the mask of the first layer is now the aplha channel and the layer itself without mask is the new picture.

[jim]

Hi Selden,

Celestia writes only stupid things to 'align.dat'.

Code: Select all

Opened registry key
Using joystick: Microsoft PC-joystick driver
nStars: 112523
.
.

 ...

Initializing ARB vertex programs . . .
Loading ARB vertex program: shaders/diffuse_arb.vp
Loading ARB vertex program: shaders/specular_arb.vp
Loading ARB vertex program: shaders/haze_arb.vp
Loading ARB vertex program: shaders/bumpdiffuse_arb.vp
Loading ARB vertex program: shaders/bumphaze_arb.vp
Loading ARB vertex program: shaders/shadowtex_arb.vp
Loading ARB vertex program: shaders/diffuse_texoff_arb.vp
Loading ARB vertex program: shaders/rings_arb.vp
Loading ARB vertex program: shaders/ringshadow_arb.vp
Loading ARB vertex program: shaders/night_arb.vp
Loading ARB vertex program: shaders/glossmap_arb.vp
All ARB vertex programs loaded successfully.
render path: 5
.
.
.
 
 ...

Saving preferences . . .
Opened registry key


This should be no problem of keyboard because there is no difference in this case. I work still with Win98se and maybe this is the reason (like to save Cel-URLs).

Another thing, I build a new billboard for textures in 2X x X format (e.g. 2048x1024). I think this is sometimes very usefull for rectangular textures to reduce the memory need. If you want I would send this to you for your Billboard-site.

Bye Jens

[selden]

Jens,

Paul specified M104 in "galaxies.dat" with two definitions because of its shape. One part is an elliptical galaxy and the other is a spiral. The Messier galaxies catalog is a copy of those original definitions.

Sure: please send a copy of your rectangular "billboard.

I suspect you're right about the Win98 keyboard problems. I don't know why it wouldn't pass those keystrokes to the running program. This suggests that there are other key combinations that don't work, too.

Is there any chance the "record angles" keyboard command problem is due to a European keyboard mapping? Some accent marks (like ~, which enables the console log window) require you to type a space afterward. "!" is a special case in Spanish, I think, because there's also the upside-down version of that character.

[selden]

Jens,

I'm confused by your description of an alpha channel problem. I've never seen anything that matches what you seem to be saying.

Could you provide a couple of pictures illustrating the problem? One showing it and one with the fix?

Thanks.

[jim]

Hi Selden,

I will try to illustrate the problem a bit. The first picture shows what I mean with my "equation".



This compare shows that the unfixed version appears to dark and dark dedails are vanishing.



The example may not be the best with a red nebula this is much more dramatic.

Bye Jens

PS: "... rectangular textures ..." this should be "rectangle textures"

[selden]

Jens,

I think now I understand the problem you're fixing. In fact I mention this problem on my
billboard Web page.

To describe it in different words,

When you use an alpha transparency channel with a picture of a "Nebula", the transparency channel makes the object look darker than it should be. The regions that the alpha channel makes partially transparent also are made proportionately darker. A gamma correction can be applied to the picture to compensate -- to make it proportionately brighter again where the alpha transparency channel would darken it.

Have I understood this correctly?

P.S. I've included a link to Jens' rectangular billboard on my billboard page. It was just a quick addition. I hope to do a major rewrite of the whole page in the not too distant future.

P.P.S. Jens, "rectangular" is fine. It means "shaped like a rectangle".

[jim]

Hi Selden,

Have I understood this correctly?

It's sometime hard for me to descripe with my poor english what I mean but now you got it.
I've also tested to apply a simple gamma correction but this makes the colours very pale.

Minewhile I found a more simple way to add transparency to the nebula texture.
- Change the colour mode to CMYK,
- copy the black channel to alpha channel,
- and then fill the black channel white.
That's all. The result is relative good but you can apply a delicate gamma- and saturation correction to make it perfect.

Only to make this instuctions complete:
- Before save the texture set the colour mode back to RGB.
- To save it to DDS (PSD with plugin) reduce the layer to background.
- To save it to PNG add a mask with the aplha information to the layer.

Selden another topic, I think we should Chris ask if he can "hard code" a billboard mesh in Celestia that allows to use virtual textures. This would increase the perfomance with nebula immense. Further I could build a 8k version for M104. It would be also very helpful if "lowres/medres/hires" texture load work.

Bye Jens

[selden]

Jens,

Yes, I agree that it would be great if the features that work for planet surface textures also worked for Nebula textures.

I'd like the AltSurface menu to work, too. I have several Nebula Addons that I've been working on that could use it. They're just a mess, otherwise.

Could you post the Feature Request to the Users Forum? Just in case Chris doesn't see this, and so he knows it's more than just me who wants these things!

[jim]

Hi Selden,

The problem with 'align.log' is solved. It was my fault. I foregot to scroll down to the end of 'align.log' where the values can be found.

Could you post the Feature Request to the Users Forum?

I can do that but I was not very successful with my last Feature Request.

Bye Jens

[selden]

Jens,

Thanks for letting us know that it wasn't a keyboard problem!

I'm sure Chris doesn't have as much time to work on Celestia as he'd like. Hopefully knowing that more than one person wants a particular set of features will encourage him to work on them sooner.

[jim]

Hi Selden,

I think there is a bug in align edit mode. Control-Shift Left-Mouse-Button and Control-Shift Right-Mouse-Button seems to work in different coordinate systems. I noticed this while trying to align the Andromeda billboard. I'm loosing permanent the z-alignment.
Did you notice similar effects?

Bye Jens

[danielj]

I don?t like very much this add-on.M81 and NGC 4314 or the pie plate galaxy are objects with some sense of depth.You can rotate and see the two sides.But in this case,it?s a photo pasted in Celestia screen.Why it not seem a 3d model?It?s totally unrealible unless you get too close to the galaxy.I think this model could be improved.And M31,too,is a cut image very poor made.I don?t understand why the galaxies models are degrading...

[selden]

Jens,

I don't think I see the problem you mention, but I no longer use the Left-Mouse-Button rotations for billboards. I've always found it very hard to use.

When aligning billboards, they need to be face-on toward the solar system. Grant Hutchison wrote a spreadsheed calculator that provides the correct Axis and Angle values which do that.
See http://www.lns.cornell.edu/~seb/celestia/hutchison/dsc-spreadsheet.html

Then, while watching from the Earth, I use the RightMouseButton to rotate the model around the axis that points out of the screen.

Does this clarify anything?

[selden]

Daniel,

What do you mean "degrading"?
What other galaxy models are you comparing them to?

The only other 3D galaxy model that I know about is the one that I did for NGC 4622. (That one is much too bumpy, and since Celestia doesn't depth-sort the faces of 3D Nebula models correctly, I haven't spent a lot of time trying to make others.)

Jens is only just starting to learn how to project pictures onto models and how to orient them in space. I'm sure his models will improve as he gets more experience.

You might want to consider making some, too.

[jim]

Hi danielj,

I really like all these flat billboard addons and I prefer these because I want a realistic looking Celestia.

Do you think a clumpsy 3d model looks better, a model build without any real 3d information, equited with a quite fictional texture? And all this effords only to make such unrealistic things like fly around galaxy.

When do you think in reallity we can see such a galaxy from another site in 100 years? or 1000 years? or never?

Now I'm a fan of a Celestia that shows mostly the known facts and from all this deep space objects we have at the moment only this flat pictures.

But nobody prefent you from building your own addon.

No ovence!
Jens

P.S. Selden, I will try Grant spreadsheed calculator.