Fractal System Creation Algorithm

[Ortolan]

Would it be possible to produce cloud cover on the fly using fractals?

[Nikola]

I also wanted to make a system generator for Celestia, but didn't know how to make it realistic, having in mind newly discovered planets.
Perhaps planet sizes could be decided just randomly, but what about their orbits? Should they follow Thicius-Bode's law? Etc, etc...

If anyone needs some help, I'd be glad to.

[Axel]

Hi Marc,

hmmm, albeit your code had to calculate half an hour for your two first shown fractal planetary surfaces still they look very promising and convincing!

So here's my two cents:


Doppler-detected extrasolar planetary objects:

I'd use your code to create virtual extrasolar planetary *surface textures* right now. I guess it's safe to say everyone knows we have some probs with C and probably won't be able to obtain scientific data on this for the period of everyone's life here in the forum. The reflected light on a planet is pretty dim compared to it's star.

So the suggestion to prevent Celestia from becoming 'dirty' by not scientific data has no implication on this. Who does not want this can simply switch back to a wireframe model of every large planetary object ouside Neptune's orbit and almost all minor planetary objects. That's already available.

I'd prefer an ficticious, albeit more realistic presentation of extrasolar planetary surfaces in an asthetic application as Celestia for these object properties not available in a scientific manner.

To solve the problem of downloadable texture size vs. calculation time I'd suggest to implement your algorythm with limited iterations (say 5 or so) and let Celestia create these textures locally during one night if needed.


Regarding completely ficticious planetary systems:

I'd say it's pretty realistic to find the same planetary distribution we did around this and close-by stars further out. So once again a ficticious system is more realistic than an empty.

But here I agree with the fact that we should not repeat the mistakes early geographers did with unknwon spots. Fake them and present them as real.

Once again it ough to be a feature which can be disabled. If it's enabled ficticious orbits and planet names simply ought to be displayed in a diff color.

Good work

CU,
Axel

[Marc Griffith]

Axel, I agree with you on the point of not getting celestia 'dirty'. The work ive been doing is for my mostlyharmless project (a Celestia hack) it will contain Fantasy elements. I think celestia should focus on what we know.
Just to clarify, the fractal planet generator is not my work. Mark Santesson wrote it http://www.circuitpsyche.com/fracplan/

Back to StarGen http://www.eldacur.com/~brons/NerdCorne ... arGen.html

Ive managed to modify it to produce randomly generated moons as well. I dont know if they are 'accurate' but they are 'believable'. Much of the code was already there, it just wasnt being used.
The next step is to modify the text output of Stargen to so it is in ssc file format.

[Axel]

Hi Marc,

I think you got me wrong by about 180°

I don't have probs with fictional planetary surfaces where approbiate / distinguishable and I don't have probs with fictional orbital data if clearly distinguishable from real data.

Of course booth features should be to enable/diable throughout the application. And I think the large calculation time for the ractal texture generation is no prob if you do it cummulated during one night.

Hope I have been able to clearify my point,
Axel

[Marc Griffith]

I think ive got the misunderstanding.
Your talking about generating systems on the fly while celestia is running or integrating the fractal stuff with celestia? Or just a single application to do all the generation? Am I on the right track?

Ill try to describe a bit better where im heading.
Im planning on generating all the systems fractally (and randomly) first then running celestia with that data. This will include ssc files and textures. A number of applications will be used, the fractal planet generator, a gas giant generator (by Franz-Xaver Schweiger
), a modified stargen, a random name generator and some custom applications to link them all together.
To keep the size down the textures will have to be re-used througout. For instance there may be 20 different 'mars like' textures and 20 different 'ice world textures' etc.
The whole process will probably total in days of processor time. Plus a large amount of manual work.
To this I would like to add some the work that other people have been doing creating systems, Rassilon, D.Edwards and Manuel David. And also make use of some of the great textures these people have been creating.
The end product will be a huge Celestia addon, which ill probably have to buy cable in order to upload. :?

Like you said, to keep the size down it would be nice to have the end user doing some of the generation. Such that the user downloads the starter package then runs a program to generate the remaining files. How to go about this has its own problems which may or may not be achievable and is something to tackle after the bloated addon is finsihed.

[Axel]

Hi Marc,

I think ive got the misunderstanding.
Your talking about generating systems on the fly while celestia is running or integrating the fractal stuff with celestia? Or just a single application to do all the generation? Am I on the right track?

Nope, I thought at the time it might be a good idea to generate the data (textures/system/orbital data) when first needed compeletly on the user's machine. That way no large textures are downloaded, no computation time is spend for ppl who don't use the feature.

If you use it, Celestia will start external tools as needed to generate the data during one long initial over-night-session. The resulting data ought to include 100 or so fractal textures per planet type as needed by the as well generated orbital data. I wouldn't really recommend to use a very limited set of pre-generated textures as this will get repetitive very soon. You can still use predefined fictional textures, you just have to explicitly define the object and therefor exclude it from the fractal generation.

Since the fractal algorythm is deterministic under static parameters everyone using the feature would end up with the same universe after the over-night-session has been completed.

Ill try to describe a bit better where im heading.
Im planning on generating all the systems fractally (and randomly) first then running celestia with that data. This will include ssc files and textures. A number of applications will be used, the fractal planet generator, a gas giant generator (by Franz-Xaver Schweiger
), a modified stargen, a random name generator and some custom applications to link them all together.
To keep the size down the textures will have to be re-used througout. For instance there may be 20 different 'mars like' textures and 20 different 'ice world textures' etc.

Here I'd disagree. If you reasonable limit the amout of iterations of the texture algorythm you might get it to generate the 100 or so textures per planet type in some hours. You can even enhance the 'value' of the textures you just created if you COMBINE e.g. two of them to the textures finally used by Celestia.

E.g.: You have a plain Mars-type red dust texture and you have a moon like texture. If you overlay them in a nice, organic way you get a dust planet with some small caters.
Next example: You could predifine a few geological features (large rifts, mountain chains...) to enhance you standard fractal textures this way.

The whole process will probably total in days of processor time. Plus a large amount of manual work.
To this I would like to add some the work that other people have been doing creating systems, Rassilon, D.Edwards and Manuel David. And also make use of some of the great textures these people have been creating.
The end product will be a huge Celestia addon, which ill probably have to buy cable in order to upload.

Like you said, to keep the size down it would be nice to have the end user doing some of the generation. Such that the user downloads the starter package then runs a program to generate the remaining files. How to go about this has its own problems which may or may not be achievable and is something to tackle after the bloated addon is finsihed.

Hmmm, if you think to generate the 'bloated' texture/data pack why not start on the remote client right away? Distribute the code not the data.

CU,
Axel

[marc]

Looks like my original account had resurrected itself (yay)

Good points all of them, its a lot of work though.

Celestia will start external tools as needed

Integrating this into Celestia would require alot of work, Ill be sticking to external apps i think.

Hmmm, if you think to generate the 'bloated' texture/data pack why not start on the remote client right away? Distribute the code not the data. :)

Your right here. Once i get it working Ill ask the authors of code I've pinched, then Ill post the work as part of the MH project.

[Axel]

Hi Marc,

it sounds like a hell of a lot of work. But following Braben's footsteps has never been easy.

Not even for himself

Take care and keep us updated,
Axel