Page 1 of 1

Feature Request: AutoMagnitude for Deep Space Objects

Posted: 10.03.2004, 12:52
by TimMc
Following a recent thread in the scripting forum, I submit the following FR:

I hope that a feature can be added to Celestia in the future that will allow for ... limited viewing of [deep-space objects] (like the automag flag for stars).


This feature would only show DSO's that are near/bright enough to warrant displaying when moving the FOV. At present all DSO's in the FOV are rendered, even if they are so far in the distance as to effectively be point-like. As some DSO's can be very large textures on very big 3D models, the consequence can be to slow down the frame-rate to a crawl. The alternative is to switch the "galaxies" renderflag off, which may not be desirable if, for example, one is viewing DSO's. So a variable rendering facility like automag for stars would be nice.

Tim Mc

Re: Feature Request: AutoMagnitude for Deep Space Objects

Posted: 10.03.2004, 13:10
by t00fri
TimMc wrote:Following a recent thread in the scripting forum, I submit the following FR:

I hope that a feature can be added to Celestia in the future that will allow for searching of deep-space objects loaded and for limited viewing of them (like the automag flag for stars).

This feature would only show DSO's that are near/bright enough to warrant displaying when moving the FOV. At present all DSO's in the FOV are rendered, even if they are so far in the distance as to effectively be point-like. As some DSO's can be very large textures on on very big 3D models, the consequence can be to slow down the frame-rate to a crawl. The alternative is to switch the "galaxies" renderflag off, which may not be desirable if, for example, one is viewing DSO's. So a variable rendering facility like automag for stars would be nice.

Tim Mc


When I coded the automag scheme, I thought of something like this already. Yet it is harder to quantify first of all, since the brightness (in magnitudes) of extended objects is often not a relevant criterion for their actual visibility.

The automag scheme is based on (an idealized) equal distribution of many (pointlike) light sources of varying brightness.

The guiding principle was to keep their visible density constant, when the field of view (FOV) is varied. So when e.g. the FOV gets smaller many stars drop out of the field. By increasing the mag limit, more stars will become visible, however, in the narrowed FoV. The automag scheme balances these to effects against each other.

An important prerequisite to make this work according to its design idea, is that the star data base is large enough, i.e. really comprises also many weak stars down to m=14-16, say. So the extended add-on db with 2 million stars is great for it!

In case of extended DSO's there are some major qualitative differences. Strong inhomogeneities (galaxy clusters...), the extendedness of DSO's and the request to have really many such objects included are not easy to account for.

However, I agree, it might be well worth thinking more about extending the automag idea to them.

If there is interest, I would have much more to say here...

Bye Fridger

More on Graduated Rendering of DSO's

Posted: 10.03.2004, 13:19
by TimMc
Fridger

I see the problem for the purposes of accurate rendering. My main concern is to be able to move FOV with the "galaxies" renderflag activated and lots of large texture DSO's loaded without having the frame rate crawl to a halt. The FR is for scripting and aesthetic purposes rather than for pure accuracy. But such a facility would be helpful for coding educational scripts.

Regards, Tim Mc

Re: More on Graduated Rendering of DSO's

Posted: 10.03.2004, 13:24
by t00fri
TimMc wrote:Fridger

I see the problem for the purposes of accurate rendering. My main concern is to be able to move FOV with the "galaxies" renderflag activated and lots of large texture DSO's loaded without having the frame rate crawl to a halt. The FR is for scripting and aesthetic purposes rather than for pure accuracy. But such a facility would be helpful for coding educational scripts.

Regards, Tim Mc


That is more a problem of culling in the code. In this respect I had achieved already a big performance factor (10 - 30!) in the code 1 1/2 years ago...I have struggled quite a bit to further improve the situation, but with little success so far.

Bye Fridger

Graduated rendering of DSO's

Posted: 10.03.2004, 13:31
by TimMc
Hmmm. Maybe some rumination might lead to a way forward. I will leave it to you and others who have greater minds than me on such issues. Thanks for considering it.

Regards, Tim Mc

Re: More on Graduated Rendering of DSO's

Posted: 10.03.2004, 22:42
by jim
TimMc wrote:... and lots of large texture DSO's loaded without having the frame rate crawl to a halt.


That's the reson why I've stoped building new DSO's and removed round 50 DSO's from my extra folder. I was no more able to run most of this beautiful script. Until Celestia handle this not better I will not use textured DSO's. :evil:

Bye Jens

Posted: 12.03.2004, 03:47
by don
I don't know how DSOs are specified. Do they have a brightness setting? Celestia obviously knows how far away they are, for it's distance display. So why not calculate whether or not to display a DSO based on it's distance and/or brightness setting?

Forgive me if this is a stupid question. As I said, I don't know how DSOs are specified. :oops:

-Don G.

Posted: 12.03.2004, 11:08
by selden
Don,

Currently there are three types of deep space objects definable in Celestia (using DSC catalogs): OpenCluster, Nebula and Galaxy.
Brightness (AppMag or AbsMag) is not one of the descriptors available to describe them :(

A name, orientation, and location in space can be declared for all of them; that's all an OpenCluster definition can have. A Nebula can have a Mesh, which can be omitted. Instead of a Mesh, a Galaxy can have a Type and Detail level, which can be omitted. That's it.

But yes, it'd be nice if a brightness value could be specified, with its visibility controlled by the Magnitude Threshold controls, just as stars' are.

.

Posted: 12.03.2004, 20:40
by don
Thank you for the clarification Selden.

selden wrote:But yes, it'd be nice if a brightness value could be specified, with its visibility controlled by the Magnitude Threshold controls, just as stars' are.

Yes, I think this would be great for increasing FPS, when there are many DSOs in the user's extras directory.

Posted: 12.03.2004, 21:47
by maxim
...and lots of large texture DSO's loaded without having the frame rate crawl to a halt.


Wouldn't it be a solution to allow external definition of LOD-maps per DSO? In other words: something like a VT for DSO's. It's unlikely that you watch more than one DSO from closeup.

maxim :?: