Accretion disk (and magnetic field) technique

[Cham]

I'm experimenting a new technique for nice fully 3D accretion disks, without any sharp edges (for all viewing angles !). It really looks like a volumetric blurry disk, from the outside :



I'm using 72 tilted planes with a 5 degrees increment step for their EquatorAscendingNode. The SSC code looks like this :

Code: Select all

"Accretion disk 1" "PSR J0142+61"
{
   Mesh "planar1.3ds"
   Radius 1.8E6
   Emissive true

   EllipticalOrbit {
      Period               1
      SemiMajorAxis        1E-10
      Eccentricity         0
      Inclination          1
   }

   Obliquity      5
   RotationPeriod       9
   EquatorAscendingNode   0
   Albedo               0.01
}

"Accretion disk 2" "PSR J0142+61"
{
   Mesh "planar1.3ds"
   Radius 1.8E6
   Emissive true

   EllipticalOrbit {
      Period               1
      SemiMajorAxis        1E-10
      Eccentricity         0
      Inclination          1
   }

   Obliquity      5
   RotationPeriod       9
   RotationOffset      45
   EquatorAscendingNode   5
   Albedo               0.01
}

... etc ...



The result is very good as seen from the exterior (no sharp edges visible). However, there are lots of clipping plane problems from the inside.

I'm not sure this experiment will gives something really usefull, since it makes the video card working extremelly hard (I could hear the computer's fan getting very loud). The frame rate was about 15 FPS only

Of course, I already tried with less planes, but the result was much less convincing...

[buggs_moran]

Very nice Cham, look forward to your further experimentation. Have you tried randomizing their rotation speeds for gradual changes?

[Cham]

Differential velocity was my next step, to simulate a real fluid. But I'll abandon this experiment : it's too heavy on the video card !

I'm trying to find another solution right now.

[Johaen]

it's too heavy on the video card !

I'm curious. What video card are you using? I'd be curious to try this out on my machine (the one in my sig), because nothing in Celestia has really taxed my video card at all. Celestia seems to be mostly limited by my CPU.

[Cham]

I'm using an ATI Radeon x850 XT video card, with 256 MB of video ram. My machine is a dual G5 2 GHz, with 4 GB ram.

Frame rate is low, because there are 72 planes with a 1024 X 1024 PNG texture applied to each of them.

[Cham]

Here's my PSR J0142+61 model, with its accretion disk and strong magnetic field :

[chris]

It would be nice to add some basic volume rendering to Celestia--a ray marching algorithm with volume textures is a much more efficient technique for doing these sorts of effects.

--Chris

[Cham]

If only we could have fuzzy models (fuzzy spheres, planes and cones). For example : a plane with a fuzzy thickness.

[buggs_moran]

It would be nice to add some basic volume rendering to Celestia--a ray marching algorithm with volume textures is a much more efficient technique for doing these sorts of effects.

--Chris

Oh yes, please think on that Chris... The uses for this are too numerous to count. I would put my vote in for the ability to volumetrically fill a cmod if that's possible, therefore allowing us limitless shapes. (I am of course personally thinking of matter transfer streams). When I saw that atmosphere hugging an asteroid some time back I got excited about that possibility. Would there be serious depth sorting issues with a volumetric fill around, in front of or behind other cmods?

[ANDREA]

Here's my PSR J0142+61 model, with its accretion disk and strong magnetic field :

Cham, your new accretion disk improvement is very very nice and credible.
Just a little suggestion: I think that the cones are too much sharp and luminous, so for my shows I have notably reduced their luminosity, contrast and sharpness, and IMHO this way they look more "real".
Bye

Andrea

[Cham]

Andrea,

you may be right : the jets are too sharp. I'll test with more transparency textures.

[Cham]

Here's a test release of the pulsar PSR J0142+61 (updated).

http://nho.ohn.free.fr/celestia/Cham/PSR_J0142.zip (4.8 MB zip file)

Take note that the pulsar itself is a part of Selden's pulsars addon, so if you already have it, it will probably conflict with this release (you'll have the same pulsar rendered twice, but mine is set at a different distance from Sol). Just disactivate temporarily Selden's addon to test the above addon.

[Cham]

I'm now experimenting another idea for the magnetic field representation. Using Mathematica, I created the magnetic "enveloppes" around the pulsar, that match perfectly the lines. I'll try some color variations, from vivid red inside to weak green outside :



It's pretty, in wireframe mode :

[Cham]

Ooo, this is a pure mathematical beauty. I must do a CMOD version of the wireframe mode :

[Cham]

... and I would love to show the radiation pattern. Hmmm, that's a project for the far away future I guess ...

[Cham]

Here's a variation of the same dipolar field, without emissivity and without the field lines :

[Cham]

I've updated the download link to include the magnetic "layers" :

http://nho.ohn.free.fr/celestia/Cham/PSR_J0142.zip (4.8 MB zip file)

Feedback would be appreciated.

[Johaen]

I've updated the download link to include the magnetic "layers" :

http://nho.ohn.free.fr/celestia/Cham/PSR_J0142.zip (4.8 MB zip file)

Feedback would be appreciated.

After seeing the pictures I wasn't sure if I'd like the "magnetic layers" as you put it. But seeing it in Celestia is amazing. They add just a hint of a shell, and I think it looks really nice. I don't know how accurate it is, but it sure looks good.

[Cham]

I'm now experimenting some distorded magnetic field lines. I finally found a way to do it properly. The pure dipolar field from the pulsar (below, a dummy sphere for testing purposes) is distorded by the plasma located in the accretion disk. The hot plasma is a conductor, and its rotation around the pulsar builds up a large electric current which adds its own magnetic field. Physically, this problem is highly non-linear since the plasma is also in interaction with the pulsar's dipolar field, so it's a non-trivial matter to find the real TOTAL magnetic field. I will not attempt to solve the full Maxwell-Magneto-hydrodynamical equations. I'll make a simple crude representation only.



Here's a QT movie showing the field lines of a realistic model, made on a supercomputer (sorry, I lost the reference !).

http://nho.ohn.free.fr/celestia/Cham/Divers/t300.mov (2.5 QT movie)