Celestia Forums

I'd like to propose a planetary surface view (for Earth to start) for people who like to "land" on planets. I find when I "land" on Earth the textures distort the view, but I noticed something I liked a lot; atmosphere. Even at night the sky colors would be different from "in-space", and you could even add a realilistic view of the atmosphere when leaving it into space. The fading of sky color for example would look really good.

You could add an extra set of controls specifically for surface based viewing; maybe a latitude/longitude control in case you want to switch modes and go to a certain place on Earth (looking up). Who wouldn't want to see what the night (and even the daytime) sky is like from Spirit's Gustev Crater on Mars?

interesting idea.

Thanks. I know just about every other "planetarium" starbrowser is a land based magnifier, but none would be able to match what Celestia could do for the genre; i.e. DIFFERENT PLANETS/Moons, actual consideration for surface location on planet/moon, location in the cosmos, realilistic atmosphere, atmospheric entry/exit, etc. etc.

I welcome more comments; kinda curious about that one guy who thought it wasn't a good idea.

Starman,

I'm confused. Except for displaying an alt-az coordinate system, my impression is that most of what you're asking for already exists in Celestia. It'd help my understanding, at least, if you could make explicit "bullet items" for each of the features you don't see in Celestia.

Q/A#19 in the "preliminary User's FAQ" at the top of the Users Forum lists the "planetarium" features currently available.

Hi Selden,

Okay, that cleared a few things up for me. Thanks for linking me to those functions; the viewpoint may be stationed 628 meters off the sea level, but that still works for me.

The still relavent part of the question is, are you ever going to make the surface textures (especially on Earth, Luna, Mars) more physically reliefed? For example;

Could I be able to sit at the base of Mt. Everest, and have it block my view?

More interestingly, can I sit at the bottom of Gustev Crater and see the crater lip taller then me besides the sky?

I guess what I'm asking for is some ground to land on.

I'm not a programmer or an artist, so forgive my ignorance if that is at all impossible. I just think relief would add even more enjoyable realism (Can you imagine landing on the top of Olympus Mons, and looking down as well as up?). I really love Celestia the way it is; what you guys have done and are doing is very cool .

People occasionally have discussed what it might take in order to add some form of real vertical relief to spherical bodies, but nothing has come of it yet. Of course, doing it with reasonable accuracy for the whole Earth would need a rather large database. More than a TeraByte, I suspect.

individual locations can be modelled, however. It's a several step procedure, but one can use a 3D modelling package and convert DEM (Digital Elevation Model) terrain information into a 3DS model that Celestia can use.

At least one person has mentioned doing this for Valles Marineris on Mars, and Jestr has provided a model of Stonehenge at http://www.geocities.com/jestrjestr/extras/extras.html While it isn't really a terrain model, it does show what can be done with current tools.

The steps are easy to describe and sound fairly simple, but the devil is in the details.

Specifically:
1. locate (or create) a DEM model of the area you're interested in
2. convert it to 3DS format
3. locate (or create) a surface image of the coloration of that region
4. drape the image over the model
5. position the result at the right place on the planet

There are many Web sites devoted to doing these kinds of things, except for the 5th step. Most people are interested in looking at the various regions on their own.

Starman wrote:I'd like to propose a planetary surface view (for Earth to start) for people who like to "land" on planets. I find when I "land" on Earth the textures distort the view, but I noticed something I liked a lot; atmosphere. Even at night the sky colors would be different from "in-space", and you could even add a realilistic view of the atmosphere when leaving it into space. The fading of sky color for example would look really good.

You could add an extra set of controls specifically for surface based viewing; maybe a latitude/longitude control in case you want to switch modes and go to a certain place on Earth (looking up). Who wouldn't want to see what the night (and even the daytime) sky is like from Spirit's Gustev Crater on Mars?

Starman,

I am amazed that 12 people of 13 find your "idea" good (I have not even voted;-)). I just doubt that a fair fraction of the 12 has completely understood what you tried to suggest above. Neither have I.

What you write above, for me is quite an unclear mix of features, most of which Celestia offers already. It is, for example, not at all clear to what extent you are aware of Celestia's present potential in the direction you were alluding to. Could it just be that a number of your suggestions would simply be answered, and we could spare the polling, if you'd go once more attentively through the Celestia manual....?

Bye Fridger

I think that Starman is referring -among other things- to some kind of real-time displacement. It could be achieved using -for example- the normal map or bump map as a displacement map. This way, the surface features that now are volume-faked using bump/normal maps will be truely 3D modelled into the planet shape (there will be real protrusions and indentions). Perhaps it could be done using Virtual Textures, with the modelling detail of the surface as a function of viewer proximity to it, in order to save resources. This will allow the surface features to effectively block the user's view (as Starman asks for the Everest).
One problem with it is that the most common bumpmap/normal map textures are equal or less than 1km per pixel side, so you indeed will see the Everest in low resolution (it will not show much detail), but less prominent features will be barely recognizable. One solution is to use more detailed DEM (or another format) maps for certain areas of interest, in a similar way to the current VT albedo/color maps. There are elevation maps up to 10 meters resolution available on the internet.

Other solution could be a mixture of both techniques: a low/medium resolution displacement map enhaced in real time with a higher resolution bump map over it, to give the illusion of greater detail.

And speaking of no real time displacement: about a year ago I tried to define Mars as a 3DS object. Using some low res MOLA data I displaced a planar mesh and then, using scripted functions, converted it into a sphere. Although the 65565 vertex limit per mesh of my 3D modeller program, the most prominent surface elements were clearly recognizable. Then I saved it and updated solarsys.ssc, but -probably due to some internal differences in the 3DS format, or Celestia code limitation- it wasn't rendered correctly.

Starman wrote:... kinda curious about that one guy who thought it wasn't a good idea.

Perhaps [s]he was just lodging a protest against a poll that seems to be forcing the answer you want to see. It's of course quite difficult for anyone to justify saying that your idea is a bad one, but many people might feel it's a pretty neutral sort of concept that they could do without, or perhaps they would prefer the developers to spend their time on other things instead. But by omitting those sort of middle-of-the-road options, your pretty much obliging people to choose "good" or (like Fridger and me) to not vote at all - which therefore ensures an apparent vote in your favour.

Grant

About displacement maps, some graphics cards(mainly, Matrox Parhelia and ATI's latest Radeons) have hardware displacement mapping features. It would be kind of cool to have a seperate version of Celestia for each type of graphics card(I know this would be a bitch, so don't even take this as a request, just wishful thinking)...so that things like this would be feasible.

A down to 'Earth' view of the intricate detail of a celestial object would add a new dimension to this simulation. I cannot begin to try to do this. Someone please deliver!

Ive configured the model of the whole earth from the Space Graphics site to work in Celestia and even though it is not particularly detailed (just bumps for mountains really),it is a struggle to move around it in Celestia.I think it will be some time before our computers and graphics cards would be able to handle the detail.And anyway at the moment you cannot use virtual textures on a model.I think it is better for now to use VT's.Though I suppose you could model small bits of the globe and place them in the right spot.Jestr

For goodness sake, I didn't think the idea would invoke this much controversy or response. Sorry I've haven't posted lately, but my computer's down, and I'm using a friend's. I think most of you get the general idea. No, I didn't know about the "planetarium" mode, but forgive me for not reading the manual.

I did propose the 3-D relief surface idea. I began with this post trying to think of a way to acually land on the surface of an planet without having the bitmap fluxuate (I don't know if its actually the bitmap, but the colors would be all out of wack, in the sky, black ground, etc, whenever you would try to get to ground level.). Then I thought about how cool it would be if I could land on a hill, look up at the stars, and then look down again at a city or crater below. No fluxuations, but actual up-close detailed surface features that I could go right beside (or inside?).

It was just an idea. I'm sorry if my poll made anybody feel like I wanted you to vote for me or against me, or anything like that. I just wanted to know if you thought it was a good/bad idea. The reason why I singled out that one guy who thought it was a bad idea was because I really did want to know what he thought (and yes, I admit I was proud of my idea).

I just thought relief would be cool.

--Starman

Starman wrote:For goodness sake, I didn't think the idea would invoke this much controversy or response. Sorry I've haven't posted lately, but my computer's down, and I'm using a friend's. I think most of you get the general idea. No, I didn't know about the "planetarium" mode, but forgive me for not reading the manual.

I did propose the 3-D relief surface idea. I began with this post trying to think of a way to acually land on the surface of an planet without having the bitmap fluxuate (I don't know if its actually the bitmap, but the colors would be all out of wack, in the sky, black ground, etc, whenever you would try to get to ground level.). Then I thought about how cool it would be if I could land on a hill, look up at the stars, and then look down again at a city or crater below. No fluxuations, but actual up-close detailed surface features that I could go right beside (or inside?).

It was just an idea. I'm sorry if my poll made anybody feel like I wanted you to vote for me or against me, or anything like that. I just wanted to know if you thought it was a good/bad idea. The reason why I singled out that one guy who thought it was a bad idea was because I really did want to know what he thought (and yes, I admit I was proud of my idea).

I just thought relief would be cool.

--Starman

Starman,

the issue of adding a 3D-surface relief feature, we discussed many times before you joined in. You should easily be able to trace the discussions using the Search engine. Nothing new here.

The basic question remains: from where does one get the required detailed surface information? You just argue it would be "cool". Well usually the Celestia design philosophy is somewhat more ambitious;-)

Celestia is a most precise real time space simulation and not a space game of some sort where one just pops up landscapes out of the blue (e.g. by means of a random generator).

So how did you imagine to provide the landscape information around the town where you grew up?;-)


Bye Fridger

I would like to remind all 3D terrain lovers of this site:
http://www.skylinesoft.com/corporate/corporate_home.asp
There is a complete Mars model available now.
(It requires a new version of the viewer; if you have an older version installed, you have to deinstall it first).
Try the customer showcases too!

I wouldn't mind seeing a terrain feature in celestia- perhaps if someone actually made one and I tried it out, I wouldn't think it was great- but just imagining it right now it seems like a good idea.

Fridger wrote:The basic question remains: from where does one get the required detailed surface information?

Well, it would require a fair bit of modification to Celestia, but here's one way you could make the think work...

We already have Virtual Textures, which can display items as small as your house on the surface of a planet. These also work for normalmaps, so how about using a normalmaps to give the big picture of terrain height, and then for more interesting sections (eg Mount everest or something) you could go to higher levels of the VT normalmap, and then upon landing next to it, you'd see it represented in 3D. (Obviously you'd use the standard texture VT's to provide surface detail for these terrain sections)

The fact that you're using VT's would also mean that you're not constantly loading *huge* DEM's etc- and also you wouldn't need two seperate systems for shading and terrain.

Fridger wrote:You just argue it would be "cool". Well usually the Celestia design philosophy is somewhat more ambitious;-)

Heh. We already have a good bit of eye candy stuff- to be honest this isn't such a massive leap, to show surface detail with a bit more accuracy, now is it?

Fridger wrote:Celestia is a most precise real time space simulation and not a space game of some sort where one just pops up landscapes out of the blue (e.g. by means of a random generator).

I wouldn't like a random generator, as it takes away from Celestia's simulation of reality. But if someone wanted to make an "artistic" terrain for a portion of some other planet, i'd have no problems downloading that.

Good idea/bad idea? Something to think about at least!

btw- I didn't vote in the poll, but only because of the way the question was phrased. I just think that planetary surfaces should be integrated into the rest of the program the same way as everything else, there shouldn't be a special mode for it.

An afterthought... In my idea- there should be some outer limit to when 3D terrain will be rendered- only when the terrain is protruding whatever part of a degree above the horizon, or something like that. Not a set limit anyway, since it would screw up extreme terrain, like you have on Miranda with cliffs 11km high... (Now THERE's something i'd like to see! )

JackHiggins wrote:btw- I didn't vote in the poll, but only because of the way the question was phrased. I just think that planetary surfaces should be integrated into the rest of the program the same way as everything else, there shouldn't be a special mode for it.
An afterthought... In my idea- there should be some outer limit to when 3D terrain will be rendered- only when the terrain is protruding whatever part of a degree above the horizon, or something like that. Not a set limit anyway, since it would screw up extreme terrain, like you have on Miranda with cliffs 11km high... (Now THERE's something i'd like to see! )

I didn't vote, for many the already quoted reasons... As regards the 3d surfaces modelisation, the beautiful attempts by Rassilon and Jestr for particular locations are admirable and works well. But the same system for all celestial objects in Celestia does not seem yet possible for the moment, unless turning the program very very hard, yes? And there are few objects of which we have a knowledge in 3d supplements... Just Earth, the Moon, and Mars?
Jeam (Now THERE's something i'd like to see! ) ?videmment

JackHiggins wrote:I wouldn't mind seeing a terrain feature in celestia- perhaps if someone actually made one and I tried it out, I wouldn't think it was great- but just imagining it right now it seems like a good idea.

Fridger wrote:The basic question remains: from where does one get the required detailed surface information?

Jack,

of course, I would tremendously enjoy a sophisticated 3D terrain display. Yet, after all this work that led to a most accurate incorporation of the movements of celestial bodies (remember hundreds of terms in VSOP87 ...) and all the rest, I'd consider it a bad mistake to mix in 3d landscape fiction, even if it may be switched off with a button.

The point is that with Celestia getting increasingly known in the world, it also stands more and more for a great and unique concept: the blending of state of the art precision simulation with exciting 3d graphics! As soon as one starts violating this basic philosophy even slightly, Celestia will start loosing its reputation (no matter whether things can be switched off or not!).

Yes, I would also download somebody's 3d landscape code for reasons of curiosity. But sure enough, I would loose interest in Celestia very soon after...

So far the Celestia code has nowhere become inconsequent as to its inherent philosophy of accuracy in simulation. Should we really want to give it up like this??

For similar reasons, I have always been against a default integration of rendered galaxies & nebulae into the core distribution. Despite great individual success (Rass'), we have not managed so far to come across a mass production algorithm that would allow to render not one, or two or perhaps 10 but all 10000 members of the NGC catalogue! Also we are seriously missing 3d information in that case, which sabotages again the intrinsic Celestia philosophy.

So like in some very famous paintings, I think it is often the superior compromise to only sketch/outline the faces of people instead of painting them explicitly but badly!

Bye Fridger

We already have Virtual Textures, which can display items as small as your house on the surface of a planet. These also work for normalmaps, so how about using a normalmaps to give the big picture of terrain height, and then for more interesting sections (eg Mount everest or something) you could go to higher levels of the VT normalmap, and then upon landing next to it, you'd see it represented in 3D. (Obviously you'd use the standard texture VT's to provide surface detail for these terrain sections)

The fact that you're using VT's would also mean that you're not constantly loading *huge* DEM's etc- and also you wouldn't need two seperate systems for shading and terrain.

Well, this is -more or less- what I pointed out in my above post.
In principle, a complete modeling of the planet could be possible using a Virtual Textures strategy (with the appropriate code modifications of course).
The idea is that the 'displacement map VT' is used to displace the earth mesh, replacing faked bumps with real volumes.
Getting even closer, the usual 'dm VT' tile splitting process begins, so only the on screen (ie visible to the viewer) part of earth is displaced, saving memory and frames per second.

As JackHiggins implied there could be some sort of roughness/radius-relative proximity setting: when the user is closer to the object than it, the 'dm VT' starts to displace the earth mesh, replacing faked bumps with real volumes.

The standard (color/albedo) VT will cover the displaced mesh as usual, giving color to the elevation data.
All would work in a continuous manner, without 'space' or 'surface' modes.

As for the origin of data, as I already said (and I am sure that some of you already know) there is a large amount of DEM, DTED, SRTM and NED files on the internet. Some of them are of incredible high resolution (10 meters/pixel) More common ones are of 30-60-100 meters/pixel (enormous detail, indeed).
Some free utilities as 3DEM can seamlessly stitch together this files to build larger surfaces. Some others as Wilbur can make normal maps with them, change projections, flood fill them, etc.. Most utilities including the quoted ones can export terrains as tga and other graphical formats.


Fridger:

I completely agree with your post as a statement. But why do you refer to 3D terrain display as fiction? Is that a general assertion? Or are you pointing to specific feature proposals?
In my particular case, I am talking about a 3D engine using real elevation data from satellital sources, in an absolutely analogous way to the one which Celestia has been using from a long time with bump/normal maps.
I am not supporting things like random noise functions, procedural trees and things like these, but accurate scientific data rendering.

Technical issues apart: can't 3D surface volumes be achieved this way without hurting Celestia philosophy?