Celestia Forums

Hi everyone. I've always been disappointed in how early-type (elliptical and lenticular) galaxies are rendered in Celestia. Instead of looking like purplish-gray lumps, they are supposed to look like this:


In general, elliptical galaxies are very smooth, featureless, and yellowish in color. I was able to take advantage of the simplicity and create a realistic elliptical galaxy model, and how CMOD files work, to make a decent-ish model.

In CMOD files you can define polygon meshes with a bunch of properties, including diffuse and opacity. However, if you don't define a texture, Celestia just shows the outline of the model. You can take advantage of this by changing the diffuse color to match that of an elliptical galaxy, as well as changing the opacity of the outline. Just put a bunch of nearly transparent "shells" within each other, and voilà, you've got something looking like an elliptical galaxy!


A few things to note:

1. Performance. Each model has 100 materials (basically the "shells"), 14760 vertices, and 27200 primitives. I could make the details finer by adding more shells, faces, etc. but that would impact performance while only barely improving the appearance. As they are, they probably don't impact performance for most computers (my computer is pretty terrible), but if it does, let me know.

2. Brightness. These models are defined as nebulae, so you can't increase the brightness like you can with regular galaxies. Therefore, these models do not attempt to show absolute brightness.

3. Surface brightness. The surface brightness profiles of elliptical galaxies follow de Vaucouleurs' profile, which states that there is a linear relationship between the fourth root of the radius and the surface brightness (in a logarithmic system, like magnitudes per arcsecond squared). Now, when Fridger implemented elliptical galaxies in celestia.Sci, he used pixel brightness as a proxy to model the surface brightness (see this test). I did that too. So I'm reasonably confident that these models obey some form of the de Vaucouleurs' profile, but probably not the exact profile of each galaxy.

4. Colors. For the colors, I took median total B-V and V-R color indices from Buta et al. (1994) and Buta et al. (1995) respectively, and put them into FarGetaNik's color calculator. For elliptical galaxies, I used morphological type -5, which corresponds to E. For lenticular galaxies, I used morphological type -2, which corresponds to S0^0. There are actually intermediate classifications with intermediate colors, but I have not experimented with that yet.

5. Shapes. Currently, elliptical galaxies use the same semi-axis ratios defined in default Celestia. Lenticular galaxies are (arbitrarily) defined as disks that are 20% as thick as they are wide. The true shapes of elliptical galaxies are much more complicated; many of them are triaxial ellipsoids. Getting the semi-axis ratios and the inclinations would be difficult to do on a large scale, though, so I won't attempt to do that for now.

6. Orientations. These models are designed such that it's pretty easy to replace the galaxy models with them. For any galaxy, just use the same Axis value but invert the Angle value.

7. Minor features. Early-type galaxies are more than just boring ellipses. Some of them have shells (see ESO 381-12). Also, many elliptical galaxies aren't even elliptical, technically speaking. Their isophotes (basically contour lines but for surface brightness) may trace out a more boxy, or a more flattened (i.e. disky) shape. I am a total novice at Blender so I have no idea how to obtain these shapes.


8. Bars. Some lenticular galaxies are barred. I have a barred lenticular galaxy model below (called SB0.cmod), but unfortunately it doesn't really represent the "TIE-fighter shape" that you see in many of them (see NGC 936 for an example).

Anyway, those are about all the important things I could think of. Attached is are the models, as well as an example DSC file (). Enjoy:

I hope that with some refinement and improvement, it can be intergated into Celestia proper

Here are some fun uses of this elliptical galaxy model.

On the Motherlode I have an add-on for Hoag's Object, but the central core looks a bit off. In reality it's supposed to look like an elliptical galaxy, but right now it's got that same lumpy texture as the ring.

So I edited out the central core out of the CustomTemplate, and added an elliptical galaxy model (flattened, though I don't know the exact axis ratios) for the central core.

I'd say it looks pretty nice. On the left is the old version, and on the right is the new version. The new version will be on the Motherlode shortly.



A similar thing could be done with other Hoag-type galaxies. NGC 6028 and UGC 4599 are two examples.

Added after 5 hours 43 minutes:
If anybody is interested, here is the script that I used to generate these files.

They have one command line argument. You can add the specific morphological type you want to create, out of the following: E0, E1, E2, E3, E4, E5, E6, E7 (not coded into Celestia), S0, and SB0 (not coded into Celestia). The command-line syntax is like this, using E0 as an example:

Here's an "add-on", I guess. Below are files that replaces all the elliptical and lenticular galaxies in Celestia with an elliptical galaxy model. To install, replace the old "galaxies.dsc" file with the new one (keep a backup of the old one!), move the file "blank.png" to the models directory, and put the rest of the file in the extras directory.

Unfortunately downloading them causes pretty serious performance issues (i.e. lag, etc.). This makes sense, since these elliptical galaxy models are a ridiculously inefficient way to draw them. I've downsized the models so they have only fifty shells, and each shell has hundreds of vertices and triangles, even though they don't appear in the final picture. Plus I'd imagine it's very intensive to try to calculate opacity values for each shell, especially near the inside, where there are close to 50 layers.

It looks like you reimplemented .sci approach to render elliptical galaxies.

can be mixed xd

I tried to make a one galaxy xd

Daly wrote:I tried to make a one galaxy xd

it would be better if luke showed you how to change colors and brighten it
so you get get it brighter and a creamy white as in the pic of the real thing

other then that NICE JOB

Added after 9 minutes 29 seconds:
i installed it and its not working

do i need the DS_STORE files if so do i put them in textures or models?

trappistplanets wrote:it would be better if luke showed you how to change colors and brighten it
so you get get it brighter and a creamy white as in the pic of the real thing

The colors are derived from color indices from two papers (Buta et al. (1994) and Buta et al. (1995)); I transformed them using FarGetaNik's color calculator. So as far as I know, they're accurate, but if you have more specific critique, then please let me know.

trappistplanets wrote:do i need the DS_STORE files if so do i put them in textures or models?

macOS automatically creates .DS_STORE files for folders. You don't need them for the add-on.

LukeCEL wrote:macOS automatically creates .DS_STORE files for folders. You don't need them for the add-on.

then what is causing the texture errors and other junk

Hey! Would you mind sending me your flattened E0 cmod file you used for the hoag's object core? I'm attempting to make a Hoag's-like galaxy but the core just looks too spherical.

Added after 1 minute 18 seconds:
This is my galaxy, by the way

The Minmus Derp wrote:Hey! Would you mind sending me your flattened E0 cmod file you used for the hoag's object core? I'm attempting to make a Hoag's-like galaxy but the core just looks too spherical.

The addon was posted on the Discord server: https://cdn.discordapp.com/attachments/5586434404 ... 74797743095888/HoagsObject.zip

Apparently it was uploaded to the Motherlode (which is why it wasn't posted here), but never appeared there.

Thank you! For some reason, the little disk thingie in the center is a bit misaligned with the ring, why is that?

The Minmus Derp wrote:Thank you! For some reason, the little disk thingie in the center is a bit misaligned with the ring, why is that?

you need to mess with the tilt of the ring or the tilt of the core itself