Welcome: Aims and Status of celestia.Sci
Page 1 of 1

Author:  t00fri [ Wed, 10-04-13, 17:51 GMT ]
Post subject:  Welcome: Aims and Status of celestia.Sci


Having been a Celestia co-author and developer for 11 years, it was sad to witness Celestia development sliding into an increasingly deep coma... Being also a theoretical Astro-Particle physicist with decades of research experience, I therefore decided 2.5 years ago to pursue my own visions about a scientific 3D visualization of the Universe. This was when I started quietly with the development of celestia.Sci.

Obviously, the .Sci acronym in celestia.Sci stands for "scientific" and is to reflect both the underlying scientific design philosopy and my personal visions as a scientist. The remaining portion of the new project name is both to remind of my longtime ties and the program's “blood relationship” with Celestia (!

While celestia.Sci is building upon Celestia's SVN code base (, it is not to be viewed as a straight continuation of Celestia and its design philosophy. It is planned to become increasingly different from Celestia with time, because its development is driven by a number of distinctly different aims.

Of course, celestia.Sci will be Open Source. In less technical terms, it means that --along with the executable program-- the full source code is made available. Open-source software is typically developed in a public, collaborative manner. According to the GNU GPLv.2 license of Celestia, Open Source is a prerequisite for using the previous code (copyrighted by the Celestia Development Team), in an independent derived software project.

After 2.5 years of secluded development, I am now very happy to announce that celestia.Sci is entering public development in a SourceForge SVN repository very shortly! Moreover, a distinct and competent “celestia.Sci Development Team” has now begun to form:

  • Dr. Fridger Schrempp (aka t00fri), project lead.

  • Da-woon Jung (aka dirkpitt aka DW) has recently joined. Being a software engineer by profession, he is not only a co-author and longtime developer of Celestia, but also has maintained the MAC-OSX-specific code base in the past. It is simply great to have him on board.

  • Christophe Campos (aka ElChristou), a professional graphics expert and old Celestian in love with Apple computers ;-), co-founder of this CelestialMatters site ...and a good friend!

At this occasion, let me express an invitation typically to interested students of astronomy and / or astrophysics with a good C++ background, to participate in the project. In return, participants will gain experience in scientific working methodology under the guidance of an experienced astro-particle Prof. who has successfully supervised a large number of PhD students in the course of time ...

Principal Aims

In comparison to Celestia, celestia.Sci will be much more devoted to visualizing Deep Space and the transitional regime to Cosmology.

At such large distance scales of deep space, astronomical / astrophysical data increasingly provide information of statistical nature, with data coverage being far from complete. Correspondingly, missing data can and are being supplemented in celestia.Sci by the good theoretical knowledge we have, along with a sophisticated statistical generation of objects in the Universe, such that available scientific catalog data are nevertheless perfectly well reproduced. This more theory-based approach is most interesting and instructive and leaves plenty of room for actual research questions in the course of code development!

A crucial cornerstone in celestia.Sci is my new generalized concept of color for Deep Space Objects (DSOs):

The colors of galaxies or globular clusters, for example, represent crucial scientific information about the evolution history of these objects. On the other hand, comparing any two colored photos of the same galaxy, will often reveal huge color differences, that may lead people to infer that color is NOT a well-defined degree-of-freedom in (deep-space) astronomy.

But this problematics is just analogous to the issue of color management in computing. Here --as a remedy-- one starts with measured color profiles for the monitor (in a standardized format) that subsequently allow transformations in color space to the color gamut of attached printer devices, web cams, HD TV sets etc, such that --on each device-- the image looks precisely identical!

While in Celestia, a restriction to visual perception was a basic design requirement, this scientific color and multi-wavelength approach in celestia.Sci represents a crucial new ingredient. The aim is to work out a set of standard .icc or icm color profiles that allow transforming between the color filter environment of e.g. the Hubble instrument and that corresponding to the SDSS surveys etc.. But this stuff is complex in detail and still requires some work. At present, only the equivalent of an SDSS color profile is operational and used throughout.

By default, celestia.Sci features a display of the great photographic MilkyWay (MW) imaging by Nick Risinger. It involves a precisely aligned and star-cleaned texture over a transparent sphere mesh, that is centered on the Sun. This marvellous view with its pink starforming patches of ionized atomic hydrogen (H II), induced a promising start of a new approach to so-called 'local sphere' environments in deep space. OpenGL experiments about rendering the Universe inside the local sphere, on its surface and outside of it --along with a clever way of blended 'transit'-- are under way. A next step is a number of such local spheres, each referring to the local environments of different prominent galaxies, along with the possibility of travelling between them...

Some crucial advantages of such an approach are near at hand: The local spheres approach serves first of all as a means of embedding 2D photographical images and real or statistically generated multi-million 2D star data as local backgrounds in a true 3D Universe simulation.

With regard to cosmological extensions of celestia.Sci, local spheres will play another distinguished role by separating space-time into 'flat' local regions and far ones where General Relativity effects are manifest (lensing etc)!

Galaxies and other DSOs look much more realistic with (nameless) statistically generated faint stars around. There is a promising scheme within my good old Automag mechanism from Celestia, for implementing such faint stars, far distant from the realm of the real star data. This is also work in progress.

I have picked these few aims, since they are already under concrete study. But they are just examples of many more interesting tasks, such as on-the-fly loading and unloading of specialized data catalogs, specific textures, plugins and add-ons in general. Visualization of lensing phenomina in the cosmological realm, as presently under study by Da-woon Jung, is another prominent plan. Soon or later, I plan to implement the rather huge official DSO catalog from the (Sloan) SDSS deep space survey, with lacking data supplemented by statistically generated objects. This will lead us to visualizing the famous large-scale galaxy filaments and many more unique deep space characteristics.

Already Implemented New Highlights

Graphical User Interface (Qt4): For more details, please refer to this CM thread

Here I have implemented quite a number of convenient new features:

  • More toolbars with nice icons hosting sub-menus, where one may quickly choose the typically desired configurations via multiple selections. Some icons change appearance, depending on the selection within the associated sub-menu. All icons are new and carefully designed.

    By just one click on these icons, one may show or hide the entire preferred configuration of orbit displays, grids and labels, or modify the color schemes for DSOs and stars etc.

    Besides vivid colors and grayscale rendering for DSOs, there is the so-called adaptive color scheme, where --like in visual observation-- the color of DSOs appears de-saturated when they are small (weakly luminous) and gradually become fully colored during zooming, i.e. when they occupy a significant fraction of the canvas.

  • Also the preferences sub-menu has entirely been re-coded and the configuration saving has been extended.

    There is a new preferences.ui file with various new combo boxes, handy sliders etc. This file was composed with the help of the Qt-Designer that is part of the Qt-Creator IDE, which I like very much. E.g. the speed of color-adaptation may be adjusted with a slider as well.

  • There is also a somewhat fancy full-screen scheme: by hitting Shift+F11, one may quickly change to a full-screen display that is devoid of menubar and toolbars. This looks very nice! On return (via another Shift+F11 hit), precisely the previously shown toolbars are displayed again as well as the menubar. I am also playing with an auto-hiding mechanism, but that is still under study. If a full-screen display with visible menubar and toolbars is preferred, one may simply use the full-screen button of the display window instead...

  • I have re-coded the input file for qmake such that no additional building tools are required. For Linux, even a convenient 'make install' target is part of the Makefile as generated by qmake.

  • Finally, there is a nice splash screen popping up during the initial data loading process. I used a custom splash class to achieve proper placement of the displayed text output.

Stars: By default, there are now nice-looking and fast shader stars. After plenty of shader and vertex buffer tuning, I am quite happy with them now.

Galactic Globular Clusters: While they look admittedly beautiful in celestia.Sci, it is hardly apparent that this task was by far the most work-intensive chapter so far! There were months of pure astrophysical research before I was ready for coding. In order to get the subtle star colors and the luminosity distributions right for the 157 globulars, a lot of high-tech analysis with Maple was required:

First of all, here is the long list of clickable scientific papers about globular cluster data and theoretical issues that entered my underlying analysis:
Click on titles NOT on filenames!

Furthermore, here are just two extensive, well-commented Maple worksheets that should provide a good impression about the kind of underlying research work...

Here one task was to work out via least-square fits an analytical form for the luminosity function of globular clusters, with data taken from various Hubble telescope measurements. A subsequent task was to find a universal scaling prescription for the individually measured V-I Color-Magnitude-Relations(CMR), after reddening corrections were subtracted. The final aim was to obtain a simple and fast color description of ALL globulars in terms of relatively few parameters...This effort was actually successful.

In the second worksheet,

I performed a new least-square fit of the simple yet accurate and fast King profile to the brightness measurements for about 85 globulars, with the resulting best-fit parameters being directly written into the 'globulars.dsc' data file. For each globular cluster you see two plots in the worksheet: one from before the fit and one displaying the improved matching after the fit! The improvement in realism was quite dramatic altogether.

It is another experimental project to enable the observer to select and travel to ANY of the globular cluster stars for a breathtaking sky-experience.

In this context, see my CM threads Imagine the Solar System was Part of a Globular Cluster,, and Visiting Stars in Globular Clusters?

The globular stars where rendered with the same shaders as the normal stars. For the vertex buffer I used a modern VBO approach in streaming mode.

Galaxies: The rendering code is almost completely new and now entirely shader-based.

  • First of all, a scheme was developed for a sophisticated automatic coloration of galaxies, depending on their Hubble class and their catalog value of B-V color. This scheme turned out very successful, indeed. The B-V values are read in via the modified galaxies.dsc data file.

  • Notably the rendering of the elliptical and lenticular galaxies with scientific surface brightness profiles (deVaucouleurs) was a non-trivial and time-consuming affair. Have a look in the bookmarked examples for an idea about the new 'almost photo-realistic' rendering quality of elliptical galaxies.

  • In the recent CM thread celestia.Sci Test: Brightness Profile of Elliptical Galaxies,,
    I have demonstrated that the actual brightness distribution of ellipticals as measured on my monitor screen, corresponds perfectly to the logarithmic version of the celebrated deVaucouleurs R^1/4 profile for elliptical galaxies.

  • Note that I have implemented a sophisticated combined brightness adjustment ([ and ] keys) for galaxies AND normal stars. Upon pushing these keys, the galaxy brightness oscillates in a sinus manner while the stars' brightness varies monotonically. Thus, with only ONE key almost TWO independent settings may be achieved. Try it out when looking at the bookmarks!

So far so good...

Many further appealing design perspectives lie ahead of us, notably in the realm of cosmological visualization. All we need for the time to come is your support with just a little bit of enthusiasm for this young project!


(Dr. Fridger Schrempp, project lead)

Page 1 of 1 All times are UTC
Powered by phpBB® Forum Software © phpBB Group