Anyone taken a look at this ?
http://en.spaceengine.org/
Opinions ?
Offtopic
Moderators: tillias, Jennos, peteyg
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Offtopic
It looks like you're asking me to finally add ring shadows once and for all. :p
Anyway, it looks like celestia with fancier graphics.
-
- Confed Special Operative
- Posts: 317
- Joined: Mon Jan 19, 2004 12:48 pm
- Location: Konstanz, Germany
Re: Offtopic
I´m not asking anything.
I am a scientist, If I ask, I ask directly
And if I´d have a wish for free I´d wish a volumetric shader for nebulae - after a few additional coworkers for the vegatrek project, of course....
I am a scientist, If I ask, I ask directly
And if I´d have a wish for free I´d wish a volumetric shader for nebulae - after a few additional coworkers for the vegatrek project, of course....
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Offtopic
If I had a clear idea how to write that shader, I would. I'm still drawing a blank there, short of an imposter cloud, I have no idea how to raytrace the nebula fast enough.
Re: Offtopic
Not sure what effect you are trying to achieve. Glowing clouds in space?
For atmosphere clouds I've been playing with the following idea (originally developed by ATI?). You create a mesh volume for the cloud (for atmosphere clouds this volume is enclosed by two layers displaced by cloud density, think of heightmap). In your main pass you write scene depth into an extra buffer. Then you render the cloud mesh back facing tris depth into one buffer and the forward facing into another. This step can even be done in one pass using normals to differentiate between back/forward facing. In a final pass you use this three depth values to attenuate your scene.
For clouds the only bit that doesn't work is in-scattering, as you would need multiple density samples along view ray and towards light source. But for a generic volumetric fog/cloud/smoke effect the idea might be good enough.
Edit:
Hmm, for in-scattering one could do another cloud depth pass from the light direction, like with shadow mapping. And then sample it a few(3-4) times in the cloud view depth interval, interpolate density towards light. Need to experiment with it.
For atmosphere clouds I've been playing with the following idea (originally developed by ATI?). You create a mesh volume for the cloud (for atmosphere clouds this volume is enclosed by two layers displaced by cloud density, think of heightmap). In your main pass you write scene depth into an extra buffer. Then you render the cloud mesh back facing tris depth into one buffer and the forward facing into another. This step can even be done in one pass using normals to differentiate between back/forward facing. In a final pass you use this three depth values to attenuate your scene.
For clouds the only bit that doesn't work is in-scattering, as you would need multiple density samples along view ray and towards light source. But for a generic volumetric fog/cloud/smoke effect the idea might be good enough.
Edit:
Hmm, for in-scattering one could do another cloud depth pass from the light direction, like with shadow mapping. And then sample it a few(3-4) times in the cloud view depth interval, interpolate density towards light. Need to experiment with it.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Offtopic
Atmosphere clouds seen from space are already done. Depending on your GPU, the raytracing step will be more or less precise. It works pretty much like you said, albeit greatly optimized since, assuming the camera is in space, lots of things can be approximated. Like, no need to render depth, depth can be computed based on the assumption that the cloud density map is mapped onto a sphere.log0 wrote:Not sure what effect you are trying to achieve. Glowing clouds in space?
For atmosphere clouds I've been playing with the following idea (originally developed by ATI?). You create a mesh volume for the cloud (for atmosphere clouds this volume is enclosed by two layers displaced by cloud density, think of heightmap). In your main pass you write scene depth into an extra buffer. Then you render the cloud mesh back facing tris depth into one buffer and the forward facing into another. This step can even be done in one pass using normals to differentiate between back/forward facing. In a final pass you use this three depth values to attenuate your scene.
For clouds the only bit that doesn't work is in-scattering, as you would need multiple density samples along view ray and towards light source. But for a generic volumetric fog/cloud/smoke effect the idea might be good enough.
Edit:
Hmm, for in-scattering one could do another cloud depth pass from the light direction, like with shadow mapping. And then sample it a few(3-4) times in the cloud view depth interval, interpolate density towards light. Need to experiment with it.
But it's not that what I was referring to. Glowing clouds in space indeed is what I was referring to. Planetary nebula, accretion disks, artificial clouds of other kinds.
Planetary nebula don't move while flying in-system, but they would move when you jump from system to system.
Re: Offtopic
Confused... I wasn't talking about atmospheric clouds from space. But about 3D "volume" clouds that you can fly through.klauss wrote:Atmosphere clouds seen from space are already done. Depending on your GPU, the raytracing step will be more or less precise. It works pretty much like you said, albeit greatly optimized since, assuming the camera is in space, lots of things can be approximated. Like, no need to render depth, depth can be computed based on the assumption that the cloud density map is mapped onto a sphere.log0 wrote:Not sure what effect you are trying to achieve. Glowing clouds in space?
For atmosphere clouds I've been playing with the following idea (originally developed by ATI?). You create a mesh volume for the cloud (for atmosphere clouds this volume is enclosed by two layers displaced by cloud density, think of heightmap). In your main pass you write scene depth into an extra buffer. Then you render the cloud mesh back facing tris depth into one buffer and the forward facing into another. This step can even be done in one pass using normals to differentiate between back/forward facing. In a final pass you use this three depth values to attenuate your scene.
For clouds the only bit that doesn't work is in-scattering, as you would need multiple density samples along view ray and towards light source. But for a generic volumetric fog/cloud/smoke effect the idea might be good enough.
Edit:
Hmm, for in-scattering one could do another cloud depth pass from the light direction, like with shadow mapping. And then sample it a few(3-4) times in the cloud view depth interval, interpolate density towards light. Need to experiment with it.
But it's not that what I was referring to. Glowing clouds in space indeed is what I was referring to. Planetary nebula, accretion disks, artificial clouds of other kinds.
Planetary nebula don't move while flying in-system, but they would move when you jump from system to system.
PS: I don't mean this as a feature request of any kind, just sharing some info.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Offtopic
For those, in space, in-scattering is the entire effect. Nebulas, more precisely, are glowing clouds interspersed with opaque ones. The whole neat effect is produced by in-scattering and self shadowing, and that's what I don't know how to accomplish.log0 wrote:Confused... I wasn't talking about atmospheric clouds from space. But about 3D "volume" clouds that you can fly through.
PS: I don't mean this as a feature request of any kind, just sharing some info.
Dust opaque clouds are easier, but still challenging. An imposter cloud can easily handle direct lighting, and simple tricks can mimic self-shadowing without actually simulating it (ie: paricles closer to the light source aren't shadowed, particles further away are with the shadowing increasing as you go deeper). The challenge here is sort order and overdraw. They can easily kill most GPUs (see VS's leak effect on damaged ships and the effect it has on performance).
-
- Confed Special Operative
- Posts: 317
- Joined: Mon Jan 19, 2004 12:48 pm
- Location: Konstanz, Germany
Re: Offtopic
Not necessarily glowing clouds. Simple clouds would be ok for me. Just a kind of "fog" in space...Not sure what effect you are trying to achieve. Glowing clouds in space?