Celestia Forums

Yesterday, I had the fortunate of stumbling upon Celestia while looking for a good skychart/planisphere.

After having played with it for a bit, I'd like to offer a few suggestions (I'm sure most of them have been asked countless times already):

1) How about making it a bit more friendly for use as a planisphere? Additionally, is anyone working on an interface for telescopes with computer drive systems?

2) Implement a physics engine. Computational complexity would rear its ugly head and authenticity would go out the window, but it could still be amusing/educational. Very weak, limited approximations are better than nothing. It would have to be an "ugly hack" in many ways--you can't very well expect to reasonably simulate a system with thousands of bodies, each having an affect on everything else. But a modest implementation would open up all sorts of interesting possibilities.

3) Implement volumetric nebulae and other assorted voluminous features. The form and shape of their volumes doesn't need to be very precise, but anything would probably be more precise than ignoring that they exist altogether.

4) 3D surface features. I want to fly to mars, land on it, and then scurry about its surface (with an attendant gravitational approximation). Bump mapping data for various planets already exist; we could use such data to protrude the surface when you get close enough. This would increase the importance of very high resolution bump maps.

5) More small touches, like solar flares, black holes with accretion discs, etc. The more the better.

6) Support higher detail for planets etc. without requiring display adapters with more than a gigabyte of local memory. ;) As you get nearer and nearer, load new textures for those smaller regions.

Additionally, support the loading of data over networks so we don't need to have a local copy of everything (rather only a relatively small cache). A few planets have very high resolution imagery of them stored on large terabyte databases; it would be nice if we could make use of 'em.

(Btw, does Celestia support the rendering of texture data directly over an AGP bus instead of storing the texture in local texture memory first? Local memory could be used more like a cache, leaving most of the texture data in system memory.)

7) Celestia interface for external application control. Example usage: an educational web page about our Solar System could use Celestia for some interactive visuals.

That's it for now. Any thoughts, suggestions, or criticisms to add? :)

At least one of those features is already available !

Black holes with accretion discs - see Bruckner's add-on repository. The adress can be found on the Celestia home page, somewhere.

I don't know about any of the others, I decided to quit watching the prereleases of 1.25, wait for the final version (I'm still waiting !!!) and then do a massive search of the forums and get all the updates.

1a) goto earth; goto long,lat. look around. What could be simpler?
1b) nope. Why bother? center HIP #, engage magnification.

2) That's a different program for a faster computer

3, 4 & 5: Tp a certaom extemt. the 3d aspect is already available. Someone just has to make the models. Two proofs of principle are available on Rassilon's Web page http://www.shatters.net/~rassilon/: a planetary system around pulsar PSR1257+12 (the pulsar model) and the Eta Carinae Humunculus Nebula. My understanding is that Celestia does have problems with really large objects, though. Another problem is that we have very little information about the 3d shapes of most nebulae, primarily how they look from one direction and some info about how the gases move inside.. Some have suggested "billboards" so at least they look right from one viewpoint.

5) animated textures and animated 3d models are not yet supported,

6) yup.

Additionally) Isn't that already available? just use softlinks. But be prepared for your system to page itself to death. Unfortunately, Celestia requires the image information to be in one of three formats: jpeg, png or dds. tiff is a recognized file type, but I dunno where or if it actually can be used.

7) one external control module has been developed. See the thread
http://63.224.48.65/forum/viewtopic.php?t=1253

Make it easier to make custom stars and when this is done, stars should have the ability to be made in other galaxies (not to mention planets).

by the way, how do you make comets?

First, I'd like to mention another idea...perhaps impractical or beyond the scope of the probject, but I wanted to mention it anyway.

This is sort of an extension of the external application control, but it addresses the idea of manipulating Celestia in various ways using a message passing protocol. Celestia would also need to embrace an open metadata format and plugin architecture for various ways to respond to event messages.

Examples follow:

You have Celestia running on your main display. It's connected to various severs over the Internet which can pass message packets to it. One of those servers your client is connected to is a global weather server (you can specify which packets you accept and don't accept on the basis of metadata filtering if you so choose). Suddenly the weather system passes an event to your client informing it that a hurricane has formed in some region of the sea. To handle this data and visually depict the hurricane, you need to download a "hurricane simulater" Celestia plugin. The more details it passes to your client about it, the more detailed and accurate your rendering will be of it. By feeding this plugin various statistics like size, wind strength, and location, the plugin will construct a visual representation of this hurricane as accurately as it can in real-time.

You could do the same thing for things like supernovae, solar flares, volcanoe eruptions, etc. Not everything needs or warrants some graphically near photo-realistic representation; simple labels/symbols/etc would suffice for most things like earthquakes, large forrest fire outbreaks, and even things like countries/states/zip code boundaries, traffic congestion, or any other sort of information people might want to be presented with. The data doesn't necessarily need to be real-time updates of actual events, nor does it need to come from across a network; your client could accept messages from local processes as well.

I didn't do a very good job with explaining this in a more generic, flexible way nor was it very concise. If I had more time I would have put my thoughts to word more effeciently, but hopefully you get the idea. At this point I'd argue, given all of this functionality/complexity it becomes more a question about basic infastructure/protocols/standards developement with Celestia as a standards compliant application using that generalized platform, targetted to serve a specific niche. Or maybe not. Need to collect my thoughts...

My response to selden follows...

>1a) goto earth; goto long,lat. look around. What could be simpler?<

Was hoping for a bit more contextualized information to overlay my screen when in that "mode", etc. :)

>1b) nope. Why bother? center HIP #, engage magnification.<

LOL.

>2) That's a different program for a faster computer<

Not necessarily. The simulation would have to be limited/constrained to a reasonable number of objects at a time, however. Nor would it likely have any permament affects on the Universe, other than possibly superficial visuals (if that).

>3, 4 & 5: Tp a certaom extemt. the 3d aspect is already available. Someone just has to make the models. Two proofs of principle are available on Rassilon's Web page http://www.celestiaproject.net/~rassilon/: a planetary system around pulsar PSR1257+12 (the pulsar model) and the Eta Carinae Humunculus Nebula. My understanding is that Celestia does have problems with really large objects, though. Another problem is that we have very little information about the 3d shapes of most nebulae, primarily how they look from one direction and some info about how the gases move inside.. Some have suggested "billboards" so at least they look right from one viewpoint.<

Thanks for the links. I'll look into those. I was aware of the problem accurately representing nebulae might pose, but educated guesses would be sufficient IMO. (algorithmically generate most of 'em rather than handcraft, of course)

>5) animated textures and animated 3d models are not yet supported,<

So I've come to realize. Procedural textures, particle engines, etc would be a welcome addition beyond simple static sequence animations. Imagine what you could do with clouds if you had a nice dynamic cloud generator...especially if it was tied into a weather system as outlined in my ideas above. :)

>7) one external control module has been developed.<

Thanks, I'll look into that also.

astrum aspicio wrote:Yesterday, I had the fortunate of stumbling upon Celestia while looking for a good skychart/planisphere.

After having played with it for a bit, I'd like to offer a few suggestions (I'm sure most of them have been asked countless times already):

1) How about making it a bit more friendly for use as a planisphere? Additionally, is anyone working on an interface for telescopes with computer drive systems?

There are plans to make Celestia more friendly as a planisphere, including alt-azimuth rotation controls, and an easy way to set an observing site on Earth or any other body. I'm not interested in adding telescope control. Celestia is interesting because it's a 3D space simulation that allows you to travel through the solar system and beyond. Telescope control is a feature that I think works best in a program designed around Earth-based viewing.

2) Implement a physics engine. Computational complexity would rear its ugly head and authenticity would go out the window, but it could still be amusing/educational. Very weak, limited approximations are better than nothing. It would have to be an "ugly hack" in many ways--you can't very well expect to reasonably simulate a system with thousands of bodies, each having an affect on everything else. But a modest implementation would open up all sorts of interesting possibilities.

Like telescope control, a physics engine is something that I feel belongs in a different sort of program--something more along the lines of Orbiter.

3) Implement volumetric nebulae and other assorted voluminous features. The form and shape of their volumes doesn't need to be very precise, but anything would probably be more precise than ignoring that they exist altogether.

Rendering of nebulae is probably the most requested feature. I will get around to doing this eventually . . .

4) 3D surface features. I want to fly to mars, land on it, and then scurry about its surface (with an attendant gravitational approximation). Bump mapping data for various planets already exist; we could use such data to protrude the surface when you get close enough. This would increase the importance of very high resolution bump maps.

Celestia will eventually be able to do this . . . I'm not certain right now how long 'eventually' is, however.

5) More small touches, like solar flares, black holes with accretion discs, etc. The more the better.

I'm interested in developing a more general model/animation format that will make these things possible.

6) Support higher detail for planets etc. without requiring display adapters with more than a gigabyte of local memory. As you get nearer and nearer, load new textures for those smaller regions.

Additionally, support the loading of data over networks so we don't need to have a local copy of everything (rather only a relatively small cache). A few planets have very high resolution imagery of them stored on large terabyte databases; it would be nice if we could make use of 'em.

I want to do this, but haven't had time . . . Keyhole's EarthViewer does it very well. Keyhole also has a server with 7 terabytes of image and elevation data and fat pipe to deliver the data. I don't think I'll have a similar configuration anytime soon :>. But very large textures/bump maps stored on a local hard drive and loaded section-by-section to video memory are much more manageable.

(Btw, does Celestia support the rendering of texture data directly over an AGP bus instead of storing the texture in local texture memory first? Local memory could be used more like a cache, leaving most of the texture data in system memory.)

This happens automatically in the OpenGL driver, without Celestia doing anything special.

--Chris

i don't know about how much time you have... but why skip on Celestia? suggesting a feature is better suited for another program, isn't that just giving up on the whole thing? I wish i could help with it all, but i don't know a damn thing about computer programming...

It's a matter of priorities. Chris has only so much time available. There already are several programs available that do an excellent job of telescope control, for example. Or gravity simulation.