Page 4 of 4

Posted: 19.05.2007, 02:37
by chris
dirkpitt wrote:
chris wrote:There's some sort of problem with point sprites on the Mac, and I haven't been able to figure out what it is. I'm inclined to believe that it's a driver problem--at the very least, there's nothing I'm doing that should bring down a computer completely.

For me it always either 1) Crashes just Celestia, or 2) Hard-crashes the OS.
The problem only occurs on the OGL2 path and VBO's need to be enabled.
(2) happened when I also removed the texture declaration from the cmod. I saw this debug message in my system.log after I rebooted: "kernel[0]: ATIRadeon9700::wait_for_rb_space: Overflowed block waiting for FIFO space. Have 2, need 8. RBBM_STATUS 0x84116100".

I can try out a couple of other things if it'll help.


Yes, anything you can do to help debug this would be immensely appreciated. The fact that it doesn't crash when VBOs are disabled is very interesting. Is the model rendered correctly when they're turned off?

--Chris

Posted: 19.05.2007, 02:49
by dirkpitt
I don't think so. The color for example is yellow when lit, and suddenly changing to blue when part of the plume is covered in shadow.

Posted: 19.05.2007, 03:59
by volcanopele
Just to give you an idea on the true color and appearance of Io's plumes, here are a few Galileo images:

Prometheus and Zamama on limb
http://pirlwww.lpl.arizona.edu/~perry/io_images/11ISCULANN01.jpg

Zamama on limb, Prometheus against disk (plume invisible)
http://pirlwww.lpl.arizona.edu/~perry/io_images/11ISVOLUND01.jpg

Zamama against disk (you can see some brightening to the west of the lava flow) and Prometheus near terminator (note reddish plume shadow)
http://pirlwww.lpl.arizona.edu/~perry/io_images/11ISMARDUK01.jpg

Pillan plume on limb, Zamama near terminator (note brightening over plume), Prometheus beyond terminator
http://pirlwww.lpl.arizona.edu/~perry/io_images/11ISPILLAN01.jpg

Prometheus and Zamama against Io's disk. Prometheus' plume is invisible while Zamama's plume is only seen as a bluish tinting of that volcano's lava flow field
http://pirlwww.lpl.arizona.edu/~perry/io_images/14ISPOLAR_01.png

Ra Patera plume on limb
http://pirlwww.lpl.arizona.edu/~perry/io_images/G1ISPLATEN01.jpg

Masubi plume on limb (violet filter only, note classic umbrella shape of Prometheus-type plumes)
http://pirlwww.lpl.arizona.edu/~perry/io_images/22ISPLUSUR02.jpg

Posted: 19.05.2007, 09:13
by ANDREA
chris wrote:
ANDREA wrote:
chris wrote: I agree that they're too bright . . . --Chris
Chris, I installed the latest available cvs (dated 30-Apr-2007 11:26 3.5M) over Celestia 1.5.0pre2, ... and I don't see them as showN in your close-up images.
That version lacks a recent fix that corrects some scaling problems with point sprites. 1.5.0pre3 will be out soon though.
--Chris

Thank you Chris, these are good news. :wink:
Bye

Andrea :D

Posted: 20.05.2007, 01:46
by buggs_moran
Alright,

I have finally figured out my Tvashtar problems. So, here it is to play with. I will be adding a couple more volcanoes, so if there is any input, I am all ears... I am not sure about the brightness. I commented out my EmissiveTrue statements and I get a shadowy plume. I think I figured out the timing pretty close based on a paper I read that stated a 500 m/s initial velocity. I'll admit, I fudged the figures a lot. Other problems include the fact that the plume is based on circular, not parabolic motion. It is the only way I could get it to work with my idea. If it is too slow on some machines I can put out a chopped down version. Once I get some input on the look and the brightness, I will put it on the motherlode.

edit: You will need to speed up time a bit to get the full effect...and again, this needs some work, I am sure it can be cleaned up as far as the cmod and graphics go. It works fine on my system (see specs below).

download: http://www.clanmoran.net/celestia/addons/iovolcano.zip

Image

Posted: 20.05.2007, 02:05
by Hungry4info
That's a beautiful image. Thanks for the download too.

Posted: 20.05.2007, 14:02
by ANDREA
buggs_moran wrote:Alright, I have finally figured out my Tvashtar problems. So, here it is to play with. I will be adding a couple more volcanoes, so if there is any input, I am all ears... I am not sure about the brightness. ... Once I get some input on the look and the brightness, I will put it on the motherlode. download: http://www.clanmoran.net/celestia/addons/iovolcano.zip

Buggs, it looks very nice, and the motion is outstanding.
Just a couple suggestions:
1- the overall dimensions are IMO excessive, at least the double of the real plumes, as for what I've seen in the real pictures, if I haven't missed someone;
2- the moving plumes get darker when they are disappearing under the surface, but I think they should be even darker at this point, AND brighter at the beginning (the brightness difference is too small, IMHO);
3- in my system the motion is absolutely smooth and without problems, more than 60 fps (60 is my given max limit), so after checking other lower-end cards, could perhaps be possible to increase the number of "OrdXX" "Sol/Jupiter/Io/Tvashtar" and "Ejecta" "Sol/Jupiter/Io/Tvashtar/Ordzz" in ssc script, in order to better investigate the full possibilities of this system? :wink:
Very nice work, Buggs.
Bye

Andrea :D

Posted: 20.05.2007, 16:55
by buggs_moran
Actually, in my rush I made a STUPID error. It is precisely double the size it should be Andrea...

I will fix it this afternoon...

Posted: 20.05.2007, 17:24
by ANDREA
buggs_moran wrote:Actually, in my rush I made a STUPID error. It is precisely double the size it should be Andrea... I will fix it this afternoon...

Thank you Buggs. :)
BTW, I've edited the volcanoes.ssc file from Chris' point sprite cmod Io plumes, eliminating all volcanoes except the Tvashtar one, i.e the same that is in Buggs addon.
Well, even if I've not the latest Celestia compiled, this is the result, that shows how the two ways can be co-added without any problem:

Image

At this point I would like that someone with latest installed cvs will make the same thing, i.e. install both addons in extras directory (possibly AFTER Buggs will have modified his plume dimensions), substituting the original volcanoes.ssc with the following one:

Code: Select all

# Plumes of active volcanoes on Io. From J. Perry. The direction of
# increasing longitude is east (as on Earth)

# Name:                Lat   Lon

# Tvashtar Paterae      63  -123

"Tvashtar" "Sol/Jupiter/Io"
{
   Mesh "plume.cmod"
   Radius 500

   OrbitFrame { BodyFixed { Center "Sol/Jupiter/Io" } }
   BodyFrame { BodyFixed { Center "Sol/Jupiter/Io" } }
   FixedPosition [ -459.4187 -694.1060 1619.6272 ]   
   FixedRotation {
       Inclination 28.2     # 90 - latitude
            AscendingNode -33.5  # 90 + longitude
        }

   Orientation [ 90 1 0 0 ]

   Albedo 0
}

This just to see how it could appear inn the final version.
Any volunteer over there? :wink:
Thank you
Bye

Andrea :D

Posted: 21.05.2007, 00:20
by buggs_moran
It seems my site is down for some reason or another so I cannot post an updated version. Will try again in the morning...

Andrea,

I tried 72 ords instead of the 36 and things got a bit choppy, but were very interesting looking. It needs some work. I will have a little time this week, so I will try to get to it. I also don't like the shinyness and will have to take care of that.

Posted: 21.05.2007, 01:33
by buggs_moran
Download capability is back. Everything seems better.

FYI, on my laptop (onboard graphics, nothing special, no GL2) I cannot run Chris' point sprites with my cmods. Once I deleted the other set, mine worked (on April 30th build if that helps Andrea). With long, lat, radius and plume size I can add a volcano model wherever someone wants one. I plan on refining the graphics and cmod this week... I am satisfied with the clouds that eminate, but not the particle streams and shine that the plume seems to have. I am also toying with the idea of inner plumes, any thoughts?

More to come...

Posted: 21.05.2007, 03:58
by chris
buggs_moran wrote:Download capability is back. Everything seems better.

FYI, on my laptop (onboard graphics, nothing special, no GL2) I cannot run Chris' point sprites with my cmods.


So what happens when you try to use them?

--Chris

Posted: 21.05.2007, 15:48
by dirkpitt
OK, more bad news about point sprites on OS X.
This is a post I found on the mac-opengl mailing list from last week:

From: "Wesley Smith" <...>
To: macopengl <mac-opengl...>
Date: Fri, 18 May 2007 23:31:52 -0700
Subject: Point Sprites Freezing OSX
I'm rendering point sprites with coord_replace and shader
point size enabled. Don't know if that makes a difference, but it's
worth mentioning. If I render > 6000 (give or take a few thousand)
point sprites, my system seizes up and I get the beach ball. I can
move my mouse but I have no keyboard interaction nor mouse click...I'm on OSX 10.4.8 intel imac with radeon x1600....The same code I'm using on the x1600 fails miserably on the 9700 despite the extension be listed as supported.


There has been no official response from Apple yet.
Note that I also have a 9700, and reducing the number of points does not make a difference, only switching off VBO and even then the colors look wrong.

Posted: 21.05.2007, 18:20
by volcanopele
Here are a couple of views comparing buggs_moran's model (from last night) and two New Horizons views taken at the same time:

Ihires5: compare the Celestia view to the actual New Horizons image

Rripple3: compare the Celestia view to this New Horizons clip

In summary, the plume looks a bit smaller, and is darker at high phase angles that expected.

Posted: 21.05.2007, 23:44
by Hungry4info
Agreed, the plumes are dark in comparison to the New Horizons images. But remember that the New Horizons images are over-exposed to show the plumes better. So naturally, they're going to appear brighter.

By the way, this odd-on's drags frame rate on my machine. It's a nice add-on though, I just need to get a better computer.

Posted: 22.05.2007, 06:33
by SkyScraper
I have a weird problem with point sprites i see them as dots when they are textured but when i remove the texture i see them as a point sprite looks like without a texture.
I tried to change the texture format from jpg to bmp, png also in the cmod file but no efect
I have the latest CVS build installed and OpenGL2 on, i done everything right.
The addon i downloaded was http://www.celestiaproject.net/~claurel/celest ... iteneb.zip
It seems my video card is not suporting hardware sprites but as i sad early i see the sprites if i remove the texture

Can somebody tell me if should install a specific video driver or my PC configuration is not compatible with this stuff

My configuration is:
Windows XP SP2
Intel Celeron D 2.5Ghz procesor
512 DDR of RAM
Ati Radeon X700PRO video card
160 Gb Sata 2 hdd

Excuse my bad english

Posted: 22.05.2007, 07:23
by volcanopele
Hungry4info wrote:Agreed, the plumes are dark in comparison to the New Horizons images. But remember that the New Horizons images are over-exposed to show the plumes better. So naturally, they're going to appear brighter.
Very true. But the plumes still have a forward scattering phase function, and should appear brighter at higher phase angles than at lower ones. The brightness, I should note, seen in the first image I posted, is just about right. The plume just needs to be a bit bigger.

Posted: 22.05.2007, 11:36
by buggs_moran
Hungry4info: You can delete every other ejecta in the code to speed things up... As I said in an earlier post, I will be cleaning up the texture and the cmod to refine it when I get more time.

volcanopele: Yes it is too small. I need to work out the details a bit. The plume size is correct, but the placement at the surface needs to be tweaked. Hopefully I will have more time later this week.

Posted: 22.05.2007, 12:00
by selden
SkyScraper,

Install the most recent graphics driver for your video card. That might help. Drivers are available for free on the AMD/ATI Web site. Graphics drivers are updated frequently.

Posted: 09.06.2007, 02:14
by danielj
An update.
Iovolcanes are now working beautifully in Celestia 1.5.0 pre3.Maybe the new driver Forceware 94.24 had fixed the problem?