New Celestia-1.4.0pre-FT1.1 Version for Download

[Cham]

I downloaded the FT1.1 version for OS X compiled by Steven. I have some comments to say about this version and the galactic renderings.

Fridger and Toti :

Firstly : THANKS a LOT for all your efforts. It's REALLY appreciated ! The galaxies are very nice (vastly better, actually), despite some crittics to be described below.

Now about thee crittics :

1- The galaxy template used for the Milky Way is an horror ! By default, it shouldn't be that one. This barred spiral is much too simplisitc and ugly. I think the original template used was much closer to the real thing.

2- The same barred spiral as in crittic #1 is very ugly for all other barred spirals. It just doesn't look natural or accurate. In my humble opinion, this template should be redone completely. It looks distorded in some way.

3- The luminosity variation while rotating around a galaxy isn't right. There are too much variations. As seen from some angles, the galaxy may almost completely disappears from the screen ! I think there's a bug here. It's obvious to me that something is wrong in the code.

4- The central galactic blob is way too faint, too transparent for most galaxies. It should be an almost opaque blob of light. There are simply not enough of those small gray spheres (granules ? small blobs ?) in there.

Well, that's all for the moment.

[Boux]

Sh*t!
Three attempts to d/l the tar ball that stall between 90 and 97% done.
This server is a pain in the a$$!

[t00fri]

Hi all,

here is an intermediate status report about reported bugs in FT1.1:

fixed

1) Sync Orbit bug (Y) for galaxies only
2) double-labels bug (spotted by Hank)
3) improved gcc 4.x compilation (XCode 2.1 friendly)

confirmed

1) flickering and lack of rendering of deep sky add-ons under WINDOWS only. All works fine under Linux.

NOT confirmed

1) Daniels long loading times for 16k large textures
2) ElChristou's "NY from the sky at night"

Bye Fridger

[t00fri]

Cham,

just some simple comments about your comments

+++++++++++++++
1) The task that we are facing is NOT to get a few particular galaxies /perfectly right/ but rather to get 10000+ galaxies mostly right! Never forget that!
+++++++++++++++

2) Our SBc template is computer-mapped from a famous SBc galaxy!

3) The Milky Way is actually SBbc (instead of SBc right now) , which would imply more and fainter arms than SBc. At present, we are limited however to the main seven Hubble classes for spirals with subclasses coming in FT2...

4) Possible individual (custom) templates for each galaxy will be implemented in FT2, hence together with Toti's more easy-to-use vectorizer a new window for people's creativity will be opened soon.

5) A compensation of luminosity variation artefacts e.g. for edge-on view of galaxies, can only be effected for the default setting of brightness. As soon as brightness levels are changed (increased!) e.g. due to incorrectly adjusted monitors, the compensation must fail!

Bye Fridger

I downloaded the FT1.1 version for OS X compiled by Steven. I have some comments to say about this version and the galactic renderings.

Fridger and Toti :

Firstly : THANKS a LOT for all your efforts. It's REALLY appreciated ! The galaxies are very nice (vastly better, actually), despite some crittics to be described below.

Now about thee crittics :

1- The galaxy template used for the Milky Way is an horror ! By default, it shouldn't be that one. This barred spiral is much too simplisitc and ugly. I think the original template used was much closer to the real thing.

2- The same barred spiral as in crittic #1 is very ugly for all other barred spirals. It just doesn't look natural or accurate. In my humble opinion, this template should be redone completely. It looks distorded in some way.

3- The luminosity variation while rotating around a galaxy isn't right. There are too much variations. As seen from some angles, the galaxy may almost completely disappears from the screen ! I think there's a bug here. It's obvious to me that something is wrong in the code.

4- The central galactic blob is way too faint, too transparent for most galaxies. It should be an almost opaque blob of light. There are simply not enough of those small gray spheres (granules ? small blobs ?) in there.

Well, that's all for the moment.

[ElChristou]

NOT confirmed

1) Daniels long loading times for 16k large textures
2) ElChristou's "NY from the sky at night"

Forget point 2, it was just a really bad joke as you said

Things are pretty nice, I love the new dimension this catalogue gives to Celestia...
As Cham sometime I feel the templates could be better (sic ), but as you said above, I will wait the Toti tool to customize templates on FT1.2 to enter complety in this topic...
Again Tx to both for this BIG step.

[Toti]

Oh, look, NY from the sky at night!!



Ok just a funny bug, I'm sure because I'm using old templates... this afternoon I will download the total new package...

I confirm you are using old templates, and also an old catalogue (it's Fridger's first one with only one name per object)

[ElChristou]

I confirm you are using old templates, and also an old catalogue (it's Fridger's first one with only one name per object)

Yep, was my first test with only the app (quickly downloaded at Steven page) with the first templates/catalogues released... Guys, don't loose more time on this. Tx.

[Toti]

4- The central galactic blob is way too faint, too transparent for most galaxies. It should be an almost opaque blob of light. There are simply not enough of those small gray spheres (granules ? small blobs ?) in there.

We discussed this one several times already. Certainly it wouldn't be difficult to add some denser areas towards the object's center but this implies an increasing cost in performance that low end systems can't afford.
The current rendering technique gives acceptable results for the most basic hardware setup but has its own very obvious drawbacks. The best approach would be to implement different algorithms depending on the render path used; this way owners of better cards would enjoy vastly improved views.
This is of course planned.

[danielj]

It should be this.But I don??t understand very much.Look:
16k*8 bits=128 Mbytes.

Celestia FT 1.1 can??t support high resolution textures.At least 16k .JPG textures don??t work.
When I load some high resolution textures to the hires folder,Celestia starts to take ages to load.Actually,it didn??t load at all.
Maybe a 128 MB video card can??t afford 16 k non compressed textures.Is it???f so,i would like to understand why.

I just checked the loading of a 16k mars.jpg texture with FT1.1 in WINDOWS which from the point of hitting G (goto) took <20 sec to display. That is a perfectly normal goto + loading time for a SOLID (non-tiled) 16k JPG texture.

There is really no understandable reason whatsoever why FT1.1 should be different in large texture loading times, since we never touched that part of Celestia.

But of course with a graphics card that has ONLY 128MB of RAM one should better refrain from loading 16k textures. You really need 256MB for this. I recommend to use VT's for such resolutions.

Bye Fridger

[christoria]

So I downloaded and posted an error early on, then I found the solution. This version FT1.1 seems to have trouble NOT displaying labels on anything called "OpenCluster", even if labels are off. I just renamed them all to "Nebula" and it fixed itself. Anyone else get this issue?

[symaski62]

Oh, look, NY from the sky at night!!

http://nho.ohn.free.fr/celestia/preview/M84bug.jpg (IMG1)

Ok just a funny bug, I'm sure because I'm using old templates... this afternoon I will download the total new package...

That's a little bit of a BAD joke!

Here is how M84 looks in FT1.1:

http://www.celestiaproject.net/~t00fri/images/m84-cel.jpg (IMG2)


ElChristou's display engine must be ages old, since it didn't even show the alternative names of M84! M84 is a beautifully round E1 galaxy and great fun to explore at close distance...


Bye Fridger

img1 => apparent magnitude (-2.9)
img2 => ------------------------ (10.0)

^^ error Apparent magnitude panne !

[t00fri]

Oh, look, NY from the sky at night!!

http://nho.ohn.free.fr/celestia/preview/M84bug.jpg (IMG1)

Ok just a funny bug, I'm sure because I'm using old templates... this afternoon I will download the total new package...

That's a little bit of a BAD joke!

Here is how M84 looks in FT1.1:

http://www.celestiaproject.net/~t00fri/images/m84-cel.jpg (IMG2)


ElChristou's display engine must be ages old, since it didn't even show the alternative names of M84! M84 is a beautifully round E1 galaxy and great fun to explore at close distance...


Bye Fridger

img1 => apparent magnitude (-2.9)
img2 => ------------------------ (10.0)

^^ error Apparent magnitude panne !

My god WHY!?

The two images are taken from VASTLY different distances of the observer to the galaxy. That's all.

Code: Select all

Img1: Distance 69.951 Kpc  appMag -2.9
Img2: Distance 16.950 Mpc appMag 10.0


Usually we don't make such trivial errors

Bye Fridger

[symaski62]

Code: Select all

Img1: Distance 69.951 Kpc  appMag -2.9
Img2: Distance 16.950 Mpc appMag 10.0

ElChristou => error appMag -2.9

symaski62
img1 => 69.951Kpc appMag -1.9
img2 => 16.950 Mpc appMag 10.0

M 84 galaxies

[Toti]

So I downloaded and posted an error early on, then I found the solution. This version FT1.1 seems to have trouble NOT displaying labels on anything called "OpenCluster", even if labels are off. I just renamed them all to "Nebula" and it fixed itself. Anyone else get this issue?

If a certain DSO type rendering is disabled (eg galaxy) the respective labels won't be rendered.

This has been corrected (along with the double-labels bug) to match the standard behaviour: now rendering of DSOs and DSO labels are entirely independent as it ought to be.

[christoria]

My problem was, it was rendering "OpenCluster" labels even with all labels off. I'm glad it's fixed. Thank you.

[John Van Vliet]

the precompiled .exe installs just fine

Also i found a small compile bug for windows visual studio .net ( 2002)
in dsooctree.h see the files below
there is a redefination

Code: Select all

//                    dsooctree.h
// C++ Interface: dsooctree
//
// Description:
//
//
// Author: Toti <root@totibox>, (C) 2005
//
// Copyright: See COPYING file that comes with this distribution
//
//

#ifndef _DSOOCTREE_H_
#define _DSOOCTREE_H_

#include <celengine/deepskyobj.h>
#include <celengine/octree.h>


typedef DynamicOctree  <DeepSkyObject*, double> DynamicDSOOctree;
typedef StaticOctree   <DeepSkyObject*, double> DSOOctree;
typedef OctreeProcessor<DeepSkyObject*, double> DSOHandler;
         // coment this out-see below
/* ............................................................................
template<>
void DSOOctree::processVisibleObjects(DSOHandler&    processor,
                                      const Point3d& obsPosition,
                                      const Planef*  frustumPlanes,
                                      const float    limitingFactor,
                                      const double   scale) const;

template<>
void DSOOctree::processCloseObjects(DSOHandler&    processor,
                                    const Point3d& obsPosition,
                                    const double   boundingRadius,
                                    const double   scale) const;
......................................................................................
*/
#endif  // _DSOOCTREE_H_


and the same thing for staroctree.h

Code: Select all

//                staroctree.h
// C++ Interface: staroctree
//
// Description:
//
//
// Author: Toti <root@totibox>, (C) 2005
//
// Copyright: See COPYING file that comes with this distribution
//
//

#ifndef _STAROCTREE_H_
#define _STAROCTREE_H_

#include <celengine/star.h>
#include <celengine/octree.h>


typedef DynamicOctree  <Star, float> DynamicStarOctree;
typedef StaticOctree   <Star, float> StarOctree;
typedef OctreeProcessor<Star, float> StarHandler;
         // comment out below
/*  .................................................................
template<>
void StarOctree::processVisibleObjects(StarHandler&   processor,
                                       const Point3f& obsPosition,
                                       const Planef*  frustumPlanes,
                                       const float    limitingFactor,
                                       const float    scale) const;

template<>
void StarOctree::processCloseObjects(StarHandler&   processor,
                                     const Point3f& obsPosition,
                                     const float    boundingRadius,
                                     const float    scale) const;
*/  ..............................................................................
#endif  // _STAROCTREE_H_


[Toti]

John,

Yes they are superfluous redeclarations. Does VisualStudio complain? I didn't notice any GCC warnings...

Anyways thanks for spotting those. Code is cleaner now.

[John Van Vliet]

yes visiual stuido did through en error on the redef's and was easy to track down the source of sed error

However after the small change it dose build


or full size 1024x693
http://johnscelestiapage.no-ip.com/photos/youarehear.png

i still am having a prob with gcc 4.0.1 and gcc32 still throughing the same error on fedora 4

[t00fri]

Code: Select all

Img1: Distance 69.951 Kpc  appMag -2.9
Img2: Distance 16.950 Mpc appMag 10.0

ElChristou => error appMag -2.9

symaski62
img1 => 69.951Kpc appMag -1.9
img2 => 16.950 Mpc appMag 10.0

M 84 galaxies

I knew you would find this also

I changed the distance definition slightly between the OLD version that ELChristou used and FT1, FT1.1! In the OLD version the distance was defined from the observer to the BORDER of the galaxy, while in FT1.1 its from the observer to the CENTER of the galaxy! You can easily check it yourself...

Bye Fridger

[Toti]

i still am having a prob with gcc 4.0.1 and gcc32 still throughing the same error on fedora 4

What are the error messages?
Have you read this thread?
http://shatters.net/forum/viewtopic.php?t=8121

With those modifications apparently it compiles with GCC 4.0