My Globular Cluster Package for Testing

[cartrite]

I noticed something. Not sure if this was reported. This thread got pretty big very fast. I might have missed it.
Anyhow, when a cluster is moved to the edge of the screen, the outer stars that are part of the cluster are visible and some are well within the screen far from the edge. But when the core of the cluster goes off the screen, these outer stars simply disappear. This happens when I zoom in from far away or if I go to the cluster.

m4.jpg

m4-2.jpg

Sorry if this was reported already.
cartrite

[t00fri]

Sorry if this was reported already.
cartrite

Indeed, Steve,

the first one was me in my report of my code checkin in the DEV list:

-- since the spacial extension of a globular cluster typically involves stars
BEYOND it's nominal mu25 isophote radius, these stars suddenly cease to be
rendered, if the globular's center is moved sufficiently OUTSIDE the canvas.

Then there were "rediscovery" reports by Andrew and confirmed by Cham several posts above yours.

That (minor) problem obviously needs a more general consideration about objects involving a certain "fuzziness" beyond their nominal "radius".

The special problematics with globulars is that the adequate radius for this culling act (stop rendering!) is the tidal radius r_t, beyond which NO globular stars can be bound gravitationally. But r_t is usually much bigger than the physical mu25 isophote radius which characterizes the main extent of the cluster.

Fridger

[t00fri]

Very soon, I will have a patch ready, for this multiply observed minor bug that I first noticed myself in the dev-list:

-- since the spacial extension of a globular cluster typically involves stars
BEYOND it's nominal mu25 isophote radius, these stars suddenly cease to be
rendered, if the globular's center is moved sufficiently OUTSIDE the canvas.

The reason is as originally suspected. One just has to use the tidal radius rather than the physical mu25 isophote radius in the frustum test on line 9796 of render.cpp. By definition, there are NO cluster stars beyond the tidal radius.

Fridger

[ElChristou]

Is this problem the same as the sun halo? (the halo disappear abruptly when you move the sun out of the screen...)

[t00fri]

Is this problem the same as the sun halo? (the halo disappear abruptly when you move the sun out of the screen...)

I suppose so. The effect should occur, if objects have an unsharp boundary with some activity beyond their nominal radius ( which is displayed in the canvas). Halos or residual cluster stars in the tail of the distribution are such examples.

The sudden vanishing of these objects happens as a consequence of a culling test which also involves a radius parameter. At present, the culling radius and the displayed radius are identical! Whenever the object's center is moved out of the canvas further than the radius in question, rendering stops.

So the solution involves introduction of TWO radii for such objects, one that is physically defined and displayed, with the other, larger one being merely for culling purposes and describing the actual extent of the object. I was in the process of coding this when Chris L offered to implement such two radii, yesterday evening. So I am waiting for his patch.

For globulars, the culling radius has to be identified with the so-called tidal radius r_t that is DEFINED as the distance beyond which NO stars can be gravitationally bound within the cluster. So this is definitely "the ultimate extension" of a globular. The tidal radius can be calculated for every globular from our two morphological parameters in the data base:

The core radius r_c and the King concentration c. In terms of these, the tidal radius is given as

[tex]r_t = r_c \, 10^c[/tex]

Of course, r_t has still to be converted into [ly], since r_c is entered in minutes of arc ['].

Fridger

[danielj]

I checked several globular clusters and there is a long way to go,because aside from how loose or compact the clusters,they almost look the same,with the same distribution of stars and I know from astronomical images,that the clusters are very different from one another.Anyway,it?s a BEAUTIFUL WORK.Congratulations!

[t00fri]

I checked several globular clusters and there is a long way to go,because aside from how loose or compact the clusters,they almost look the same,with the same distribution of stars and I know from astronomical images,that the clusters are very different from one another.Anyway,it?s a BEAUTIFUL WORK.Congratulations!

It is NOT the same star distribution, of course.

The distributions are directly reconstructed from the measurements about the globular clusters. When you compare the clusters at the SAME image scale, you will not only see that the overall cluster sizes are very different but also the ratios between the size of the bright central core and the total size as well as the amount of decrease in the luminosity distribution for larger distances from the center.

There is NO ficticous input involved from my side.

Except the color profile! That is indeed only generic for now, taken from the great Hubble photo of M80. In a forthcoming release, the individual color profiles will also be directly reconstructed from the catalogs .

[ANDREA]

I checked several globular clusters and there is a long way to go,because aside from how loose or compact the clusters,they almost look the same,with the same distribution of stars and I know from astronomical images,that the clusters are very different from one another.Anyway,it?s a BEAUTIFUL WORK.Congratulations!

As Fridger already told you, Danielj, sorry, you are badly wrong.
And following the Italian proverb that "one image is worth many words", here a practical example.
The four globulars are shown as seen from roughly the same distance, i.e. about 1300 ly.
As you can see they are wayyyy different, in size, dimensions and central core evidence.
My little cent.
Bye

Andrea

[julesstoop]

That's a very clear example, Andrea

There seems to be something a bit strange about the globular rendering though, judging your picture, and I hope maybe Fridger can comment on this.

As far as I know - I might be wrong - individual stars in globulars should be (independent of the size of the cluster as a whole) more or less of similar (size and) brightness. Yet in your picture the different globulars, although depicted from the same distance, show a clear correlation between cluster size and visibility of individual (clumps of) stars.
To put it differently: you need to get closer to a small globular to actually resolve the individual stars.

Is this a correct assumption from my side and is this the correct behaviour from a scientific point of view?

[ANDREA]

That's a very clear example, Andrea
There seems to be something a bit strange about the globular rendering though, judging your picture, and I hope maybe Fridger can comment on this.
As far as I know - I might be wrong - individual stars in globulars should be (independent of the size of the cluster as a whole) more or less of similar (size and) brightness. Yet in your picture the different globulars, although depicted from the same distance, show a clear correlation between cluster size and visibility of individual (clumps of) stars. To put it differently: you need to get closer to a small globular to actually resolve the individual stars. Is this a correct assumption from my side and is this the correct behaviour from a scientific point of view?

You are welcome, Julesstoop, happy you like it.
Regarding your question, I'm sure that Fridger will solve your doubt.
Bye

Andrea

[chris]

That's a very clear example, Andrea

There seems to be something a bit strange about the globular rendering though, judging your picture, and I hope maybe Fridger can comment on this.

As far as I know - I might be wrong - individual stars in globulars should be (independent of the size of the cluster as a whole) more or less of similar (size and) brightness. Yet in your picture the different globulars, although depicted from the same distance, show a clear correlation between cluster size and visibility of individual (clumps of) stars.
To put it differently: you need to get closer to a small globular to actually resolve the individual stars.

Is this a correct assumption from my side and is this the correct behaviour from a scientific point of view?

Jules,

I believe that what you're seeing actually is a bug. I checked the code, and it looks like the size of the star sprites is getting scaled by the actual globular cluster radius. That shouldn't be happening. It looks like it is only a single line fix. This line:

size = size0 / 64.0f; // Initialize size of small "star" sprites

...should be changed to use a constant value instead of size0. Fridger, do you agree?

--Chris

[ajtribick]

While poking around in the add-on, I noticed a minor bug in Celestia: Greek letter substitution is not performed on .dsc objects.

I edited the globulars.dsc file so that the Bayer designations of ? Tuc (47 Tuc) and ? Cen are formatted the same as for stars, i.e. "XI Tuc" and "OME Cen" (as opposed to the original "xi Tuc" and "Omega Cen" in the globulars.dsc file). (For consistency it might be good to modify the generator to output Bayer designations in the same format as the other Bayer designations in Celestia).

While Greek letter substitution is performed in the autocomplete area, the displayed name at the top left is left in the Latin form.

[t00fri]

Jules,

I believe that what you're seeing actually is a bug. I checked the code, and it looks like the size of the star sprites is getting scaled by the actual globular cluster radius. That shouldn't be happening. It looks like it is only a single line fix. This line:

size = size0 / 64.0f; // Initialize size of small "star" sprites

...should be changed to use a constant value instead of size0. Fridger, do you agree?

--Chris

Oops, that line is a left-over from the testing stage... Yes that /physical/ starting size should be the average diameter of RedGiants or something not too different.
Because the first 128 sprites will make up for the RedGiants in my code.

For now we can just put in a sensible fixed size0 value by hand and contemplate later whether the value makes sense physically


I will look into this issue asap.

Fridger

[t00fri]

they almost look the same,with the same distribution of stars

Here is another illustration of the diversity of the globular star distributions: NGC 6397

Here is the Hubble heritage photo:


The morphological parameters are quite extreme: a tiny core radius = 0.05' and a large King concentration = 2.5. Indeed as the Hubble photo shows, this is NOT a "normal" globular look.

Here is the result from Celestia (SVN 4450):



I think it looks surprisingly similar to the actual photo.

To Daniel I can only recommend to try and look deeper into matters before writing incorrect posts.

Fridger

[chris]

I just discovered that there are at least three globulars in the catalog that are thought to be extragalactic. M 54, Terzan 7, and Terzan 8 are all associate with the Sagittarius Dwarf galaxy. Fun stuff . . .

I've found that M 31 has quite a few known globulars, but I haven't seen any catalog giving the stellar distribution parameters required by Celestia.

--Chris

[Fenerit]

Thanks without end for this, Fridger.

[cartrite]

I just discovered that there are at least three globulars in the catalog that are thought to be extragalactic. M 54, Terzan 7, and Terzan 8 are all associate with the Sagittarius Dwarf galaxy. Fun stuff . . .

I've found that M 31 has quite a few known globulars, but I haven't seen any catalog giving the stellar distribution parameters required by Celestia.

--Chris

I think I discovered something interesting regarding M 54? It seems that it may have something to do with brightening and dimming the Sagittarius Dwarf galaxy. If one toggles the shift + u to turn globulars on and off, the dwarf galaxy can be be seen brightening and dimming. A cel url.

Code: Select all

cel://Follow/Sagittarius dSph/2008-09-28T16:49:34.24231?x=AAAAAAAAWhm3QEDu5f///w&y=AAAAAACYM/xWFp3I/v///w&z=AAAAAAAAkHEwtf8pHA&ow=0.821900&ox=-0.078769&oy=0.543590&oz=0.150950&select=M 54&fov=28.866140&ts=1.000000&ltd=0&p=0&rf=2938119&lm=2048&ver=2

Something is going on here but I'm not sure what. Here is another cel url that shows Terzan 8 and Sagittarius dSph. Sagittarius dSph dims when Terzan 8 is toggled on.

Code: Select all

cel://Follow/Terzan 8/2008-09-28T16:58:46.67858?x=AAAAAAAAXvxvb6YMCA&y=AAAAAABkBVT9ashI+////w&z=AAAAAAD43hUPRyPKEg&ow=0.883508&ox=-0.122322&oy=-0.431310&oz=0.135732&select=Terzan 8&fov=28.866140&ts=1.000000&ltd=0&p=0&rf=3986695&lm=2048&ver=2

cartrite

[ElChristou]

I'm now using a nvidia powerbook and the black hole I have reported earlier is not present anymore. So apparently it's an ATI related bug. Now I do have the negative clusters already reported by other folks here...

[Fenerit]

I just discovered that there are at least three globulars in the catalog that are thought to be extragalactic. M 54, Terzan 7, and Terzan 8 are all associate with the Sagittarius Dwarf galaxy. Fun stuff . . .

I've found that M 31 has quite a few known globulars, but I haven't seen any catalog giving the stellar distribution parameters required by Celestia.

--Chris

I think I discovered something interesting regarding M 54? It seems that it may have something to do with brightening and dimming the Sagittarius Dwarf galaxy. If one toggles the shift + u to turn globulars on and off, the dwarf galaxy can be be seen brightening and dimming. A cel url.

...

Something is going on here but I'm not sure what. Here is another cel url that shows Terzan 8 and Sagittarius dSph. Sagittarius dSph dims when Terzan 8 is toggled on.

...

cartrite

Confirm this behaviour

http://fenerit.webng.com/album/Celestia/globclu.jpg

It seem "zooming" dependant. The two image from below have been captured in same position but with a single "step" out of the mouse wheel and the effect doesn't appear.

[t00fri]

Hi all,

after spending 8-9 hours/day throughout the past week in our annual international DESY Theory Workshop (with follow-up dinners and all that), I am now back at the Globulars again...

As soon as I find some spare time, I'll write more details about the significant progress I have achieved meanwhile... Notably the scale inconsistency is gone and the efficiency for generating stars is now VERY close to 100% out to the tidal radius and for the highest values of the King concentration. Here, the efficiency was way below 1% before!! I have also experimented more with a somewhat different kind of blending (see images below!). Unfortunately, the black inversion of the globular stars for certain settings (e.g. MilkyWay background) is still with us for now...

The scale inconsistency arose due to a statistical trick that I was applying in order to improve the efficiency of the random generation of stars according to King's strongly peaked luminosity distribution of globulars. While the Von Neumann acceptance-rejection method turned out to be quite efficient until about the physical radius r_mu25, in the large remaining distance until the tidal radius, the efficiency was close to zero.

That's why I had used an approximate scaling trick to circumvent this problem....

For the stars in the tails of the distribution, my scaling trick unfortunately did only respect APPROXIMATELY the EXACT upper distance limit for stars, set by the tidal radius r_t. This approximation caused the culling problem discussed earlier in this thread etc.

The solution is now very elegant and fast. I shall report about it quite soon.

Today, let me just show a few nice images with comparisons to photos, of my "revamped" blending and rendering of the globulars. Of course there are no stars out of the tidal boundary anymore.

++++++++++++++++++++++++++++++++++++++++
NOTE: all screenshots from Celestia are taken at exactly the same field of view, so you can compare the different sizes and central density profiles of the globulars.
++++++++++++++++++++++++++++++++++++++++

Omega Centauri



Here is a VERY encouraging comparison photo:



Here is a zoom image of the central parts from Hubble photo with amazing resolution:



M 13 (Hercules)

Here you see the new Celestia rendering in comparison with a color photo from a 20" Ritchey-Chr?tien telescope:



Not bad...

M 4 (Scorpius)

Well, here is what Celestia displays


Here is what I got, when looking for a M4 photo in the net
http://www.illpic.com/


M 22 (Sagittarius)



NGC 5053

Much fainter...but pretty nice



Soon there will be a code patch and more theory background...

Enjoy,
Fridger