What Celestia should be...

[Vicware]

I just want to address my fellow Celestia fans about the general direction
Celestia could take over time. When I first used Celestia, of course as a
developer I saw all kinds of potential for what Celestia could be turned into, since it's basically an open book. I think most of the people on this
forum are thinking similiarly, from what I've seen. But I've come to the
conclusion now that Celestia should remain a scientifically correct
simulator. If others want to take the core engine for this and make
entertainment out of it, they should feel free, but Chris (I believe he's the
head-genius of Celestia) has enough work creating a solid software
foundation. As a matter of fact, I think that wondering around the real
universe is a lot more interesting, educational and entertaining than any
software I've come across in 25 years.

The makers of Celestia should concentrate on improving the content
and operation of the software as is, rather than spending a lot of time
to add stuff that is more entertainment than it is simulation.

What do you think?

Vic

[bruckner]

I am of the same opinion. Being the head of the project, Chris can take Celestia in the direction he wishes; but if I was such a wonderful C++ programmer, and had the vision, energy and time to build such a program, I'd do exactly that: try to reproduce the Universe with increasing precision, and leave open the possibility for others to take on and use Celestia as a foundation for other goals (mainly gaming, but also teaching, or even tracking telescopes, as someone mentioned here not long ago!)

There are plenty of areas of improvement for Celestia, being "just" a Universe simulator. Here's a short list of my own wishes, in no particular order:

- Nebulae rendering.
- Double/triple/... stars actually orbiting; including spectroscopic binaries.
- Cepheids (variable stars). Nearly all of them are classified under a handful of stereotypes, with defined color/luminosity/radius curves, and known periods.
- Solar flares, simulated dark spots following the 11-year cycle.
- More 3D models for asteroids, or the possibility of scaling (in ssc configuration files) current ones to obtain a wider range of shapes.
- Comet tails.
- An improved flying model (this can be discussed a lot, but an interesting proposal has been made recently).
- Neutron stars, black holes with accretion disks.

As you can see, nearly all of these suggestions are "astronomical". I feel Celestia should strive to be as faithful a simulation as it can, no more and no less. And we should help make it happen, even if it's only by making suggestions...

Best regards.

Bruckner

[Spaceman Spiff]

Greetings all,

Oh good, I'm glad this has come up.

But first, homage: Chris, thanks for and congratulations on your stunning Celestia. It shows a great deal of talent, and judging from the Celestia forum it's getting the fantastic accolade it deserves, as well as giving a lot of people fun. I've been addicted to it ever since I downloaded and ran it.

I think the two great things behind Celestia's success are:
1. it can show a range of astronomical objects in a visually accurate way,
2. these objects are largely configurable in the text files by users.

Clearly, most of the forum's discussions on development (after bugs, etc) seem to have centred on this second aspect. Many people have changed or added what they could or wanted, such as our very productive chap Rassilon, and others have asked for yet more features to be added, that can also be configured.

As with Vic and Bruckner, I too want to see Celestia tend to the greatest astronomical authenticity possible. I have so far spent much time tinkering and modifying what I can in the text files. I have had some success and I think some impressive results, but I find myself blocked off by the 'hardwired' parts of Celestia's graphics. It's fine that Rasillon and others add and create new planets from fantasy universes, but I'm still trying to get the Earth to look like it really does from those NASA Apollo and Shuttle photos…!Yet, I understand Chris's original intention with Celestia was to practice/demonstrate skills with graphics. I think that should be respected, so I recognise I shouldn't assume I can burden this forum or Chris with many wide ranging or tedious requests.

The pace of Celestia development seems to be speeding up. Versions are coming out faster than I can keep up! Although Bruckner wishes to focus on just astronomical objects, I think before Celestia gets onto new and exotic objects like nebulae and variable stars, it should consolidate on the more fundamental physics behind the visuals. After the basics are truly sorted, we can expand the data.

Thus until now I have been holding off such comments like (already noted) not being able to create close binary star systems. There are other things I'd like to be able to change: there's a whole range of issues to do with 'calibrating' the brightness and colour of everything. The Sun's photosphere is so bright, the corona, granulation and prominences can't be seen. It's colour is white, not yellow - that's a myth. Even red dwarves are bright disks of pale orange-yellow light. (Incidently Chris, you are one of the few people I have seen who rightly held off saturated sun/star colours!) Planets like Earth actually appear much brighter and washed out than the present texture maps show, because most images are processed to saturate colour. Then there are contrast issues. If you 'wore shades' to reduce Earth's glare, the Moon would appear much darker than you'd think. The Earth's clouds are as white as snow, the Moon is a dark as tarmac. It would be almost in silhouette when transiting the Earth seen from Venus (through a telescope). The 'twilight' atmosphere is 'hardwired' red, so the 'full Earth' has a red atmosphere, not a blue one.

There are also problems that might be about to occur. Unless they are spotted in time, we might find ourselves painted in a corner. Binary stars are definitely a danger. I think stars, planets and moons need to be generalised as one type of object. This lets us assign luminosity to brown dwarves, so their dark sides can glow wth heat, but be illuminated on there day sides. It also allows you to give non-spherical shapes to stars (fast spinning stars like Pleione, and egg-shaped Roche lobe stars in contact binaries). If we are to have good astronomical authenticity, we'll need to find a way to deal with the dynamic range of brightnesses of stars, planets and moons. For example, a red dwarf transiting a yellow dwarf (misnomer!) will appear to our eyes as a dark disk, almost black. You might not believe me… Some kind of adjustable 'polaroid' to manipulate the exposure level (like pupil dilation) might be in order, and also an override for those who just want to see everything 'nicely' adjusted. Then there is the problem of multiple shadows and complemetary shadow colours in binary systems. If you treat this right, you can generalise it so that any planet is also illuminated by the light of another nearby moons and planets (e.g., Earthshine on the Moon). And much, much more…

I think trying to do this first will provide the greatest challenges in graphics and programming. I'm afraid I have no graphics programming background (sorry! another duff…boo hiss), but I do have a huge astronomy and space background, so I can provide a lot of advice about facts, dimensions, appearances, etc. of astronomical thingies. OK, I might have put you off by now…

Incidently, I have a problem. I wanted to post little piccies of 'what I mean' and 'before and after', but remember seeing some time ago that it wasn't a good idea to post piccies into the Forum. I don't have a web site handy like Rasillon. Any advice?

Have I gone too far? Apols for the long-winded missive.

Regs.,

Spiff.

P.s., Chris, if you are interested in my input, but want to keep the bumph/pics away from the forum, please feel free to e-mail me.

[Matt McIrvin]

Funny you should mention the "color of the sun" issue-- I was just stricken with curiosity about that and did a little Web research that yielded results all over the map... the conclusion I ultimately came it is "it depends on what you call 'white'," and that is not uniquely determined, because of the phenomenon of color constancy. You can adjust to see a variety of illuminants as white, over a pretty wide range. The designation of the sun as yellow is a convenient convention that is true for some choices of white point but not others (and, of course, the atmosphere has an effect too).

Here's an excerpt from a Usenet post I wrote about it:


Sten Odenwald of Raytheon says that sunlight is yellow, "white light"
is just a silly traditional designation, and if you think otherwise
you're just fooling yourself:

http://image.gsfc.nasa.gov/poetry/ask/a11314.html

Phil Plait, who usually has good answers about these things, is
mystified by the whole question of why we call it yellow:

http://www.badastronomy.com/bad/misc/starcolors.html

Andrew Hamilton tackles the problem with more color-science
sophistication, and says the sun is peach-pink, given some popular
choice of white point calibration:

http://casa.colorado.edu/~ajsh/colour/Tspectrum.html

Mitchell Charity gives what I believe to be the straight scoop on how
complicated the question is, and why it might not even have an answer
without further specification, even if we try to factor out the effect
of the earth's atmosphere:

http://www.vendian.org/mncharity/dir3/starcolor/sun.html

(and since what we usually do is compare the solar spectrum from
outside the atmosphere with some illuminant designed to mimic normal
daylight, maybe this exercise tells us more about Earth's atmosphere
than about the sun, anyway.)

I think that what it comes down to is that, since the sun is THE object
whose light we've evolved to use as a standard illuminant, it may be
the single visible object in the universe for which it makes the least
sense to ask what color it is. The most we can do is compare it to
some other illuminant. It's yellower than office fluorescent lighting
and bluer than a typical incandescent bulb-- that I know for sure.

An interesting question that has some bearing on the "yellow star"
designation is what determines the hues we see when we look at stars
in a dark night sky. What's the white point then? Obviously it's
not the color of the star we are looking directly at, or we could
never see them colors at all-- and even I can tell that Betelgeuse
is red. Well, I can tell it's redder than Rigel. Maybe it's still
a matter of comparison.
[/i]

[Matt McIrvin]

Oh, and I should add that (much as Spiff said) the above is of course all moot if you're close enough to see a disk, since the thing is so durned bright.

[ron]

yes i too am surprised that no one has commented on this subject before now. obviously there are many users that like to add all the fantasy stuff but there are also those like myself that are only interested in seeing things as realistically as possible. i'm affraid that having a black monolith in orbit around jupiter does nothing for me and nor does visiting a ficticious solar system. luckily the way chris has designed celestia everyone can do their own thing and add what they want.

as far as using celestia as a game engine is concerned, i don't think it makes any sense. a good gaming engine needs a fast simplistic environment to operate in. it is a waste of processor power to use an engine that is calculating things like penumbra's and sphere oblateness etc. celestia is far too exacting and would not compete with a solar system generator written specifically for gaming.

i also agree that the basic celestia core should be improved and stabalised before trying to add support for more exotic objects. the addition of eclipse shadows was a tremendous addon, as will be comet tails. Anything that furthers the accuracy and visual efects is welcomed and is in essence what attracts people to celestia in the first place.

ron

[flym]

I agree with Vic

[chris]

My vision for the future of Celestia is very much like Bruckner's--I want to simulate more astronomical phenomena with ever-increasing accuracy in each release. Comet tails, multiple star systems, neutron stars, nebula, realtivistic effects, and an expanded galaxy catalog will all be present in some future version of Celestia. Also, I'm very interested in more accurate simulation of spacecraft . . . One thing I just recently got working was importing the flight paths of Voyager 1 and 2, so that in Celestia 1.2.4 you'll be able to follow these spacecraft on their grand tours of the solar system (Shrox, do you have a Voyager model you can send me? ) Celestia 1.2.2 introduced accurate orbits for the major satellites of Jupiter, Saturn, and Uranus; for 1.2.4 I'm going to tackle Triton, Charon, Phobos, and Deimos. More and more and more accuracy . . .

Spiff raised some concerns about accuracy . . . I'm very interested in getting more input on this. Help keep me honest, and post some images in the forum for me to look at. Just try not to post anything too huge.

The issue of brightness/exposure is a big deal . . . I confess that I haven't addressed this in a consistent way, and there's no exposure parameter to tweak the appearance of objects. This is partly due to laziness, and partly because I don't want to burden Celestia users with having to mess with too many parameters in order to get the 'right' image--using Celestia shouldn't be as hard as getting a good CCD image through a telescope! When deciding how an object should be rendered, I usually compromise between reality and expectation. For instance, you'll notice that the light illuminating planets doesn't diminish as you go farther from the sun--Pluto looks as well lit as the Earth, despite receiving less than 1/1000th of the sunlight that Earth does. But how fun would it be to fly to Pluto and find nothing but a black sphere? Just as importantly, it would be unrealistic, since the human eye compensates so well for a variety of lighting coditions. So, I chose to render all planets with the same level of illumination and attribute it to the wonderfully adaptive human eye. I've had to devise lots of tricks for compressing a huge range of luminance for the limited range displayable on a monitor--the glare haloes around stars are another example.

Some of the limitations you mention about Celestia are only temporary. The hardcoded redness of the twilight atmosphere is one . . . I wanted to release a version of Celestia with atmospheres, but I didn't have time to research and determine what parameters might be useful for controlling the colors of sunrise and sunset. I'd certainly appreciate suggestions. Eclipse shadow rendering is another feature of Celestia that I eventually hope to make much more general. You're correct that it's currently rather limited--in the interest of time, I didn't spend any effort on coding for the case when there was more than one sun. It's quite a lot of work to solve this problem in a completely general way, and I wanted to release a version of Celestia where at least eclipses within our solar system would look right. I don't yet consider the eclipse shadow problem solved, and I intend to revisit it and implement a completely general shadow system once more pressing things are taken care of.

As for generalizing stars and planets, it's not really practical. Celestia stores quite a lot of data for each planet, and there are over 100,000 stars in the HIPPARCOS catalog. The size of the catalog in memory would bloat to around 25 megs if Celestia stored all the data for stars that it does for planets. I will be adding 'annotations' for stars, where you'll be able to specify additional information. These will only be allocated for stars where the extra information is known, which I believe is a pretty small fraction of the catalog. Do you think that this would be a reasonable compromise?

I don't want to sound like I'm dismissing any suggestions . . . I'm just providing some rationale, and letting it be known that I'm aware of and intend to fix a lot of Celestia's limitations. It's mostly a matter of finding the time.

--Chris

[chris]

yes i too am surprised that no one has commented on this subject before now. obviously there are many users that like to add all the fantasy stuff but there are also those like myself that are only interested in seeing things as realistically as possible. i'm affraid that having a black monolith in orbit around jupiter does nothing for me and nor does visiting a ficticious solar system. luckily the way chris has designed celestia everyone can do their own thing and add what they want.

I'm personally very interested in fictitious solar systems, but only in ones that ones that rigorously adhere to rules of celestia mechanics and planetary science. I've got a solar system creation program for Celestia that I toy with now and then. I hope to eventually get it in good enough shape to release . . . Also, I do very much enjoy looking at solar systems created by other people, even when the colors are fantastic or a large satellite lies well inside the Roche limit :>

as far as using celestia as a game engine is concerned, i don't think it makes any sense. a good gaming engine needs a fast simplistic environment to operate in. it is a waste of processor power to use an engine that is calculating things like penumbra's and sphere oblateness etc. celestia is far too exacting and would not compete with a solar system generator written specifically for gaming.

Back in November, I rearchitected Celestia somewhat so that it can be used as the engine for a game or other application. This is not something to which I want to personally devote most of my development time, but I want it to at least be possible. I do believe that Celestia has potential as a game engine--not for some fast-paced space combat game, but for slower, exploration-oriented "hard scifi" sort of game where realism is important.

--Chris

[Rassilon]

I feel that Celestia should remain as a simulation as well. I like the ideas floating around on making it sort of a game, but this is more of an educational tool, especially for me As I have forgotten most of my astronomy, I am now remembering...

Some of my requests like adding sounds to the surface of planets and adding others that some have mentioned already aren't as important as getting all the bugs fixed...Those are the most important, so I won't really care that much if its never added. One thing I would like to see is a form of sprite collision, which would keep the 'spaceship' from passing through objects... There are soo many things that can be done with this program its amazing... It takes alot of time to develop a project like this... hehe and it puts most of the standard games on the market to shame Believe me, Armada 2 and Bridge Commander to name a few have a long way to go to even begin to simulate realism like this....

Also I apologize if my fantasy solar systems corrupt the purity of Celestia. That is not my intention... We all know there are other worlds out there and in theroy, we only seem to be able to detect the rather large gas giants or brown dwarfs encircling thier distant suns... I just try to bring to life some of this with my imagination and well, having alot of time on my hands some of this in my own dreams and in well known Sci- Fi...

I think there was someone who gave an idea on the structure of the boards...Maybe this could help others in discerning that purity...

Heres the thread...

http://www.shatters.net/forum/viewtopic.php?t=116...

Thanks again Chris and all others that have contributed to this project...Its a dream come true for most of us

[Guest]

The issue of brightness/exposure is a big deal . . . I confess that I haven't addressed this in a consistent way, and there's no exposure parameter to tweak the appearance of objects. This is partly due to laziness, and partly because I don't want to burden Celestia users with having to mess with too many parameters in order to get the 'right' image--using Celestia shouldn't be as hard as getting a good CCD image through a telescope! When deciding how an object should be rendered, I usually compromise between reality and expectation. For instance, you'll notice that the light illuminating planets doesn't diminish as you go farther from the sun--Pluto looks as well lit as the Earth, despite receiving less than 1/1000th of the sunlight that Earth does. But how fun would it be to fly to Pluto and find nothing but a black sphere? Just as importantly, it would be unrealistic, since the human eye compensates so well for a variety of lighting coditions. So, I chose to render all planets with the same level of illumination and attribute it to the wonderfully adaptive human eye. I've had to devise lots of tricks for compressing a huge range of luminance for the limited range displayable on a monitor--the glare haloes around stars are another example.

In another post today I suggested a possible mechanism to cope with this problem: Simulate brightness adaptation more literally. Internally, give all objects an astronomically accurate level of brightness. In the displays, adjust the "exposure" level automatically to put the brightest object visible in the scene at a comfortable brightness level.

For things that look bigger than dots, the brightness per unit angular area (such as a pixel) would be important rather than the overall magnitude.

This would simulate things we see in real life, such as the way stars vanish completely in pictures of bright objects in space, such as nearby, sunlit planets or astronauts. The appearance of a single object of interest would be much as it is now, but you'd be able to make accurate comparisons when multiple objects are in the frame at once.

It would have to be possible to switch it off, of course, since it's useful to be able to vary the limiting magnitude manually when looking at stars. For flying around the solar system, though, I'd imagine this automatic exposure mode to be the best way to look at things. You could see how charcoal-dark the moon is in comparison to the earth, yet still be able to examine it close-up without manually adjusting anything.

[Guest]

I am all but salivating over the idea of branching celestia into a
hard-sci-fi game. I spent a year of my teenage life playing elite
[as the first vector-graphics game available for the C64].

Now with hard sci fi I do mean hard, no warp drives, no fictitious
solar systems, just sort of elite inside our own solar system.
space stations. colonies. you know.
And imagine, flying about, you could have real physics. inertia.
gravity. (I know the planets' orbits are due to gravity alone,
but it's not implemented that way (because the system would
deteriorate quickly) but rather as orbit parameters. but your
spacecraft could really be subjected to gravity and all)

Oh, and obviously, I said "branching" above; this should obviously
never be part of celestia development proper. I completely agree that
the effort there shold remain of attaining more and more astronomical
accuracy. So celestia would be sort of plugged into the game or vice versa,
and still in the middle of an elite-style game I would be awed to
witness an eclipse or a conjunction etc.

And re: no warp drive, a single player could easily skip a few months
while travelling to another planet, but maybe for a hypothetical
multiplayer mode you would have to compromise. All the same even in
online multiplaying I think it wouldn't hurt to suspend a player for a
week or two while he's travelling to Jupiter... would make you think
before embarking; time is credits.


dab

[Mikeydude750]

Yeah, I like how you can add new stuff to Celestia, but this is, and should be, a realistic simulation. However, for the people who like both, you should add a mode switcher so you can switch between realistic and ficticious.

[Serge]

I'm just discovering Celestia and I like what I see.
Factually correct space simulation is always more appealing, but it's double the fun when that (engine) could be used for great space sim games*.


*like the upcoming http://www.online-universe.net/44index.htm

[Rhys]

Hi. I downloaded an older version of Celestia yesterday. Even as I write I am already downloading the newest version. I think the other users are right, it should continue to be scientifically accurate - any fictitious or speculative add-ons should have an on/off feature as others have suggested. It would be nice, if possible, to inculde some or all of the currently discovered extrasolar planets - perhaps some sort of automatic update feature (?). The only real way of improving this program is simply MORE ! More spacecraft, more asteroids (this may already have been done, but a very accurate simulation of Eros could be created from the recent NEAR spacecraft's survey), comet tails when they're near the sun, maybe some black-holes with accretion disks (I have no idea how feasible any of this is). I suggest making a map of this galaxy as complete and accurate as possible, without comprimising speed and/or graphics. Also it would be nice if newer versions could be downloaded as an update rather than the complete files (to save time). Finally, I think this program is absolutely brilliant - keep it up, going and free !

[Rhys]

Having now fairly throughly played around with Celestia I am convinced there is one feature which needs modifications : displaying the galaxies. It would be nice if they appeared as photographic images from a distance - currently they're practically invisible, unless I turn the brightness up to full and close the curtains.... am I being too picky ?

[flym]

displaying the galaxies. It would be nice if they appeared as photographic images from a distance - currently they're practically invisible, unless I turn the brightness up to full and close the curtains.... am I being too picky ?

Galaxies do not appear as photographic images when seen from far away.
Just think about M31: it is there, 6 time bigger than the moon!; can you see it as in a photo of a school book? Not at all!
Probabily you have to find out how to turn up the brightness *and the contrast* of the whole sky :D !

The surface brilliance of an extended source does not change when you change the distance from the observer. That because the surface of the object decrease with the square of the distance and this would lead to an increase of the Brilliance (Lunimosity/Surface) BUT the Luminosity decrease exactly with the same law.
Then the appearence of galaxies have to remain the same regadless to the distance from the observer.

Instead it would be interesting if entering in a galaxy we can see an exotic starry sky. few thousands of stars could be generated randomly and used everywhere it the universe. I guess that no body will find that a sky in M31 will be identical to the sky of the other side of the Milkway or of the Magellan Clouds.

Globular Clusters would be appreciated too...

[hank]

I agree that long-exposure astronomical images have created unrealistic expectations regarding the visual appearance of extended objects. But I see no reason why Celestia's remotely-controlled camera shouldn't have a gain control! I think it would be very useful to be able to increase the brightness of the galaxies, as can already be done for faint stars.

I like the idea of generating synthetic stars for external galaxies, and also for the more remote parts of the Milky Way. Care should be taken to generate the correct star density, distribution of star types, etc. (Planets could also be synthesized.) Repeatable pseudo-randomization should be used so that the same virtual universe is presented each time.

- Hank

[Rhys]

I realize that the Universe does not look as so many gorgeous Hubble images would have us believe. I was trying to suggest that maybe a little realism should be sacrificed in this case. Normally I'd like everything to be as realistic as possible (although I DO think there should be options to alter various aspects such as background lighting, speed up time etc.), however there's no point in having a "go to object" feature if you can't see the object when you get there ! However flym's idea is definately better than mine as it would not require creating entirely new 3D models. Randomizing the stars then storing their layout is a very good idea, provided the user cannot "go to" these fake stars as they would a real one, or they're labelled in big red letters : FAKE. This would help make the galaxy brighter - maybe even to the point of visibility ! - when zoomed out. Perhaps there should also be an option to alter the brightness of the stars - adjusting brightness and contrast is irritating and time consuming (well it is on this monitor with its crazy on-screen menus) and doesn't work very well - the galaxies are just too dim.
On the subject of brightness, would it be possible to simulate the glare of a star more accurately ? Celestia's stars are all quite dim, even up close. Obviously reproducing the brightness is impossible (and stupid) but the effects shouldn't be (again, any glare effects should have an adjustable option, otherwise none of the stars would be visible in the inner solar sytem(s) - fine for viewing the solar sytem but lousy for navigation by point-and-click).
One more thing I'd like is a naked-eye level visibilty option, which would limit the stars visible to all those visible with the naked eye (I think that's all stars above magnitude 6ish). The stars would be visible according to their own magnitude and glare effects from nearby stars (which would have to be effected by planetry shadows and distance). The options menu could get pretty full but it could be quite impressive.
While I'm here, if glare effects are discarded onto the rubbish-heap of ideas, I would like at least one aspect of them to be looked into. I notice that when close to a star it appears to illuminate the "background" stars like a torchlight ! Surely less stars should be visible, not more ?

On a completely unrelated topic I am interested in making 3D models for Celestia but can't find any decent (i.e. easy to use) software - anyone able to help ?