Celestia Forums

I have had an idea for Celestias galaxies that i personally think would greatly improve them. I think that the galaxy templates in the main models file, such as the Sba template image should be what i call "Color Sensitive" Meaning that if there is a color OTHER than grayscale colors on the template image, that the exact color would appear rendered onto the galaxy model in celestia.

I personally think this would open up a new range of possible addons for celestia. We could begin making custom templates for the galaxies that would look more realistic as far as im concerned.


For example, here is a custom template file i use for M 101, it is actually just a small color photo of the galaxy..



However celestia only renders my custom template the normal way, gray arms and an orange center (i have the magnitude limit turned up to the max in this image)




My idea would actually allow the colors to be rendered from the template file to the galaxy in celestia....like in this example image that i edited. I think this would make the galaxies look more realistic if you were to create a template for a galaxy like this. I realize that this cant be done for EVERY galaxy in celestia if such a feature existed, but still it would make for nice addons.

<a href="http://s425.photobucket.com/albums/pp332/j9203/?action=view&current=101-3.jpg" target="_blank"><img src="http://i425.photobucket.com/albums/pp332/j9203/101-3.jpg" border="0" alt="Photobucket"></a>

looks like that last image didnt work--posted the wrong image code, lets try this one of how i would like my idea to look.....

I like!

Thanks! I do realize you could get the same effect by just placing a billboard model over the galaxy, but then again that would also call for more RAM usage, where as this plan would not call for extra models or textures, thus less computer strain.

Color template's for galaxies would be a very nice improvements.
It is technically feasable to program and use it in real time.
There is data available (from spectrograms, other sources) about the colors.
(Colors can also be generated from pictures.)
It's small enough to be included in celestia.
Improves appearance.

Please add it to Celestia.

duds26 wrote:Color template's for galaxies would be a very nice improvements.
It is technically feasable to program and use it in real time.
There is data available (from spectrograms, other sources) about the colors.
(Colors can also be generated from pictures.)
It's small enough to be included in celestia.
Improves appearance.

Please add it to Celestia.

My galaxy code in the Celestia distribution of course includes carefully measured (<== HUBBLE) and visually corrected generic color profiles both for spirals and elliptical galaxies, respectively. This technique allows for much more delicate and more natural colors than entering colored dots via individual templates (as proposed above)!

Here is how M101 looks like in Celestia-1.6.x:

If you can't see the delicate colors, I suggest to adjust your monitor or increase the saturation of that image a bit...

For comparison, here is a typical color composite CCD photo of M101


So I don't think the colors of Celestia's M101 are all that bad?? (Recall that photographic film/CCD's are more blue sensitive than the human eye!!)

Please remember that generally, Celestia renders according to the appearance of objects to the naked eye or a small telescope. It should be well known meanwhile that the colors of rather dim galaxies can barely be distinguished in such case.

++++++++++++++++++
The prime DSO task within Celestia's science-based design strategy is to render ALL the data information contained in the best scientific galaxy catalogs! However, for each galaxy, these catalogs only provide some global color information like B-V or B-I coefficients from spectral measurements. This information is a priori by far insufficient for reconstructing realistic individual color profiles across a galaxy. Otherwise I would have done it long ago, of course
++++++++++++++++++

Since in the official distribution we want to render 10 000 - 100 000 (!!) galaxies automatically by computer and NOT just a few by hand, such naive "ideas" as proposed above are completely unsuited.

Before making such proposals, it would have been good to have a look at my code (galaxy.cpp) to realize that the translation from template patterns to the rendered galaxy shapes and colors is all but trivial. The dots on the templates just mark the places where sprite blobs of different sizes, brightness AND color will be attached during the actual rendering process. The nice-looking appearance only arises via the interferences of those overlapping sprites.

Due to the known qualitative similarity of the relative color profiles across many spiral and elliptic galaxies, respectively, my actual strategy for implementing individual, proper colororation is as follows:

For each galaxy, one reads out the standardized global color parameters (B-V, B-I,...) from the galaxy catalogs and codes an algorithm that modifies the parameters used in the generic color profile up to now. This will allow, for example, to make certain galaxies appear more blue than others in agreement with the respective catalog data.

This strategy is being implemented in my ongoing Celestia.Sci project.

For individual and more "artistic" attempts of rendering galaxies I recommend using add-ons.

Fridger

Shot down again i see, oh well, Still thanks for taking a look at my idea

Still, there's a need for some color corrections : many of the giant elliptical galaxies out there are populated with old red stars, so these galaxies have a yellowish or reddish teint (I'm not talking about the cosmological redshift nor any Doppler effect). Currently, Celestia doesn't color these reddish galaxies correctly.

I think that the suggestion above is still valable. The code should allow the user to change the color profile in the DSC files, even if it's done manually for a few galaxies (using a "replace" or "modify" command).

This would also bring more consistency with what we currently can do for STC-stars and any SSC object.

And there is also the case of the rotation sense. In many cases, some famous galaxies doesn't have the right rotation sense in Celestia. I personally had to make a duplicate of the model files ("SBa.png", ... etc) to reverse their orientation, and edit the galaxies DSC file manually. I was interested in only a few cases (some "famous" objects with a custom billboard-addon in Celestia, to show some real galaxies to the students).

I'm not talking about a full catalog with proper colors and rotation sense (I understand Fridger's points here about the fast rendering of thousands of galaxies). However, the code should still allow the user to edit the objects he/she wants (using a "modify" command), like what we can do for STC/SSC objects. Adding a color option is highly desirable, while maintaining the standard default rendering for all the other galaxies.

Cham wrote:Still, there's a need for some color corrections : many of the giant elliptical galaxies out there are populated with old red stars, so these galaxies have a yellowish or reddish teint (I'm not talking about the cosmological redshift nor any Doppler effect). Currently, Celestia doesn't color these reddish galaxies correctly.

For the reasons you mentioned, I use a considerably more orange-red color profile for elliptical galaxies than for spirals. That was implemented as of my first galaxy release within Celestia...That peculiar coloration for ellipticals might still not be perfect, but qualitatively at least it is there!

EDIT: Ooops... after checking again for the displayed color of ellipticals, I noted that the orange-red profile has somehow gone in the latest Celestia version. So during my many tries and improvements, it must have gone lost

This is most easy to restore. Thanks Martin for noting this!

++++++++++++++++++++++++++++++++++++++++++++++++++
But individual colorations for spirals, ellipticals and globulars are
still missing. That's the main challenge we are talking about here...
++++++++++++++++++++++++++++++++++++++++++++++++++

As I explained above, my galaxies and globulars will receive improved, individual colors in the course of my Celestia.Sci project. Both for galaxies and globulars, the standard global color coefficients B-V, B-I,... are usually available in the respective catalogs and my method of implementing this color information is the same in both cases (as described in my previous post above). For the globulars there are some extra aspects though ("blue stragglers and "red giants"... )

The code should allow the user to change the color profile in the DSC files, even if it's done manually for a few galaxies (using a "replace" or "modify" command).

In principle, there won't be any need for users fiddling with color in the DSC files, since the correct galaxy color coefficients will be read out from the catalogs anyway. But sure, if someone wants to play with B-V etc, the coefficients will be in galaxies.dsc and globulars.dsc and thus may be changed by hand any time.

And there is also the case of the rotation sense.

Since I will not get involved in any attempts that involve "handicraft" for small selected subsets of DSOs, unfortunately there won't be much progress as to the sense of galaxy rotation, I am afraid. If only there was a rotation flag available in the galaxy catalogs! The only way of dealing with the problem seems comparison with photos one-by-one and subsequent manual correction. This task is NOT for me, since I am actually striving at MANY more galaxies than "just" 10 036

+++++++++++++++++++++++++++++++
For a semi-quantitative, individual implementation of color, it appears inevitable in my view to take recourse to the standardized AND readily available B-V, B-I,...color coefficients, since only this way can the strong (and unspecified) filter biases apparent in many published photos be effectively eliminated and 10000s of DSOs be rendered with individual colors!
+++++++++++++++++++++++++++++++

No doubt: Adding individual DSO coloration is highly desirable and in progress

But........ it has to be done right!

Fridger

+++++++++++++++++++++++++++++++
For a semi-quantitative, individual implementation of color, it appears inevitable in my view to take recourse to the standardized AND readily available B-V, B-I,...color coefficients, since only this way can the strong (and unspecified) filter biases apparent in many published photos be effectively eliminated and 10000s of DSOs be rendered with individual colors!
+++++++++++++++++++++++++++++++

No doubt: Adding individual DSO coloration is highly desirable and in progress

But........ it has to be done right!

That is what I was talking about having DSO with good-colors. One object = one color.
Nice to see it's already underway, will this also be added to the mainline Celestia?

Since in the official distribution we want to render 10 000 - 100 000 (!!) galaxies automatically by computer and NOT just a few by hand, such naive "ideas" as proposed above are completely unsuited.

It is less worse if you compare them with the picture/3D-model idea from a while ago.
(Use a picture!! or 3D model!!!!!! of every galaxy)

I do realize that rendering 10000s of DSO that have very detailed models and color stuff isn't going to happen.
1) We don't have detailed data about a lot of DSO's.
2) Culling and LOD, calculating if stuff is visible will eliminate a lot, Level Of Detail will also make stuff less heavy.
This is a non-trivial task now. But there is hope, in the future we will have OpenCL-capable graphic cards.

No way, that every galaxie that will be fed into celestia will be super detailed and have 10000s of those detailed DSO's!!
This just ain't going to happen. There is no way, we have such an amount of such detailed content.

I do see a few galaxies addons, up to a few dozen. that are rendered by hand and the rest by computer.
Because it's indeed completely unpractical to apply for 10000s of DSO's.

A much more detailed discussion about

Celestia.Sci: Improvements of Galaxy Rendering

may be found at Celestial Matters, here
http://forum.celestialmatters.org/viewtopic.php?t=354

Fridger