Celestia Forums

I'm new to Celestia but I have some questions, the first I'm sure has been covered, they second I'm quite sure hasn't.

1. Are there plans for a fisheye projection from Celestia. A dome is a common environment for such astronomy visualisation applications.

2. I have developed a low cost dome projection system based upon a spherical mirror, see
http://astronomy.swin.edu.au/~pbourke/p ... omemirror/
I would appreciate comments on how hard it might be to add the warping necessary for this projection....there is lots of interest in this developing. I know how to do it, I've developed the two algorithms (OpenGL), the first which is the multipass texture mathod
http://astronomy.swin.edu.au/~pbourke/p ... metexture/
and the second is the geometry warping. These two methods are the standard ways of creating fisheye, and can be modified to create the warped images needed for my spherical mirror.

This is a great suggestion! A usable fisheye projection is already implemented in Stellarium, another open-source
astronomy simulator like Celestia but designed more like a planetarium. Eventually, I think it'd be nice for both
projects to borrow the best from each other.

The fisheye projection mode seems to me much more interesting for e.g stationary, panoramic displays of the sky above us (-> e.g. Stellarium) than for Celestia. Views of surface panoramas of objects within Celestia will remain rather meager for a while to come and the density of objects in the Universe around the "Celestia observer" is typically low.

The essential applications refer to squeezing many interesting things within a solid angle close to 4*Pi into the display...To me the advantages are severely offset by the substantially increased amount of CPU time...We could well use that CPU time for plenty of other challenging 3d display tasks.

Celestia can vary it's field of view from a few arc secs out to 120 degrees, the latter corresponding already to a quite distorted viewing perspective.

As I have argued plenty of times, Celestia would really need a conformal display mode of the Universe, to make a display of phenomena on cosmological scales feasible and appealing (e.g Sloan digital sky survey data etc.).

But as I can see things, we are going to never make it there, at least given the present 'development speed'

Bye Fridger

Just in case anyone is curious, the discussion about conformal mapping is here:
http://www.shatters.net/forum/viewtopic.php?t=4724

I'd take a crack at implementing this, except noone (including me) seems to be able to say clearly how it should
look like in 3D. Selden posted a picture, but there were 2 more pages of discussion after that, that seemed to
indicate a certain wishy-washy reluctance to actually implement the proposal.

Coming back on topic, a planetarium mode if it were to be implemented, should naturally be able to be toggled
on/off, so the typical user wouldn't have to worry about wasted CPU time all the time. Not all users will be using
fish-eye projection mode of course but there will be some who'll want to take advantage of Celestia's accurate
night-sky simulation, without having to switch to a different program.

I've been using Celestia as a planetarium for ages; I added my city as a location, and a simple goto puts me in the right place. In fact, I used it to track the Machholz comet as it moved through the Hyades.

dirkpitt wrote:Just in case anyone is curious, the discussion about conformal mapping is here:
http://www.shatters.net/forum/viewtopic.php?t=4724

I'd take a crack at implementing this, except noone (including me) seems to be able to say clearly how it should
look like in 3D.
...

In fact I do pretty well, but since so far the resonance was moderate from Chris, for example, I did not care to discuss the considerable math studies I have done so far. One elegant way of 3d generalization would invoke 'quaternions' .

Such a task of implementing a 'cosmic Celestia mode' would really need a coherent JOINT study and discussion from our experts in theoretical (astro)physics (like myself, Selden, Walton,...) & 3d graphics (Chris, yourself,..) . This is not something you do and commit with a little explaining note on the developers list AS USUAL

Bye Fridger

t00fri wrote:Such a task of implementing a 'cosmic Celestia mode' would really need a coherent JOINT study and discussion from our experts in theoretical (astro)physics (like myself, Selden, Walton,...) & 3d graphics (Chris, yourself,..) . This is not something you do and commit with a little explaining note on the developers list AS USUAL

Of course. I was just thinking that maybe a quaternion demo could be coded up, because a) it's a concrete idea,
and b) it might be nice to actually visualize the idea, because nobody seems to know for sure how it should look.
Also, perhaps more on-topic, c) 360-degree projection code could be reused for fish-eye projection. But any more
discussion on this probably ought to continue on the conformal mapping thread.

dirkpitt wrote:

t00fri wrote:Such a task of implementing a 'cosmic Celestia mode' would really need a coherent JOINT study and discussion from our experts in theoretical (astro)physics (like myself, Selden, Walton,...) & 3d graphics (Chris, yourself,..) . This is not something you do and commit with a little explaining note on the developers list AS USUAL

Of course. I was just thinking that maybe a quaternion demo could be coded up, because a) it's a concrete idea,
and b) it might be nice to actually visualize the idea, because nobody seems to know for sure how it should look.
Also, perhaps more on-topic, c) 360-degree projection code could be reused for fish-eye projection. But any more
discussion on this probably ought to continue on the conformal mapping thread.

Conformal maps are "angle preserving" according to their proper definition. In practice a 'conformal' display of the Universe would allow to smoothly scan space (and time) from your 'backyard' right to the Cosmic Mickrowave Background 'border', where the light 'is switched off' (<=> only 300000 years after the Bang). Conformal mapping allows to do it in such a way as to avoid strong angular distortion. The scaling of the distance is essentially logarithmic as in 2d.

http://www.astro.princeton.edu/~mjuric/universe/all200.gif

Since there are many /practical/ aspects to consider besides pure theory, in my view, it is too early for coding! One should first discuss pros and cons of various approximately conformal standard projections, quaternions etc in their relation to the constraints imposed by the structure of Celestia. Also the SDSS digital sky survey data still are only partially analyzed. The display must be cleverly devised such as to optimally 'hide the partial lack of data'...

But honestly, after several attempts to get all this moving (also via the developer list), I am quite pessimistic ... particularly at the present time!

Perhaps one important aspect to keep in mind:

Meanwhile more and more OpenGL-based 3d astro-simulations are coming to live! The latest impressive (yet earth-based) one is Stellarium. The gap to the 3d-coding advantage of Celestia is getting /progressively/ smaller. Most people can't or don't want even to appreciate the additional amount of precision of Celestia over competing software...

With an implementation of a conformal 'cosmic mode' and the respective (astro)physics know-how that is required, Celestia could make a big step forward ahead of its competitors...But most importantly, of course, this sort of cosmic displays is of /mainstream/ interest in the scientific and educational community!

I also had vigorously advocated the concept of /'filters'/ in Celestia. Switching on virtual 'wavelength filters', would allow us to include e.g. infrared or ultraviolet imaging data and display the strong wavelength-dependent differences. Titan would be a great example or even better: our sun! That's precisely, what professional space research also exploits.

But again, the time for implementing major new /concepts/ into Celestia seems over. virtually all changes now seem to happen without previous discussion among the developers, setting up a priority list etc...They largely refer to 3d-graphics improvements here and there.

Bye Fridger

[quote="birthplace"]?????‰?µ?