Orulex 1.1 Testing request

[Artlav]

Orulex is the dynamic scenery generator for Orbiter i am developing for some time already.
Current version information and data links are there:
http://www.orbiter-forum.com/showthread.php?t=999

Alpha version from 080418:
http://orbides.1gb.ru/orbf/Orulex-080418.zip (344 Kb)
Install over 1.0.

There was some changes in the terrain code, and the way LOD is done, now it doesn't try to use up all allocated resources, but aiming instead to get the best-looking terrain with least amount of polygons used, averaging at about 50000 polygons, increasing up to 95000 when moving.
Theoretically, it should speed it up noticeably without reducing quality too much.

If you want to test it, i wanted to know, is there any performance increase compared to 1.0, if the terrain looks better/worse than before, and what parameters look/work best?

There are some new parameters that is supposed to be tuned.
You can change them either in world studio planeted tool, or in orulex.cfg.

Texturing: (TexGen in cfg) Far-Near (0) of Near-Far (1) selects if the textures are generated from lower level up of from higher level down, first should theoretically provide a smoothly increasing texture resolution, while second is expected to give faster generation at a price of temporary blank squares.
Which one looks better is one of the testing questions.

RangeFactor: Defines the level LOD fuzziness, the higher it is, the further on the hi-res textures are generated, but the lower is their maximum resolution.

TextureFactor (cfg only): Amount of texture blocks wanting that switches the generation from low-intensity mode to high-intensity, only applicable in single-threaded mode and in world studio.

 

[fort]

Bonjour Artlav,

I would like to have more time to test Orulex but currently...
I will be on holiday (if I can say...) in ten days. I however made
some approaches these last weeks to try to appreciate what that could
return on a computer like mine of small capacity (celeron 1.1 ghz,
ATI 7500 64 mb AGP 2 (or 4?) , 500 Mo sdram). I thus discover,
or rather, I rediscover, having tested the first(s)release one year ago.
Like those of Meshland.

It is rather astonishing and the progress are very clear as for the returned than
for the lightness in terms of graphic load and CPU.

I would thus return there with a little more experiment when I could
handle all that.

Just this for today: the behavior in hover with Vcollision.dll
(computerex), on the basis of the ground of Mars, is "brutal". I know
that all this is not completed and that you also seek to develop a
system managing the collisions. But for the hour, and whoever would be
interested to know, Orulex and Vcollision does not do good housework,
it seems.

For the remainder, thank you Computerex and thank you to you for
your research.

Good day.

fort

 

[Enjo]

Thanks for bumping the thread, Fort

Lunar Suicide is a great scenario for testing Orulex, because it's independent of your controls.
First I tried the latest version and then 1.0. What struck me at first was that it seemed that compared to the latest, 1.0 didn't manage to keep up with the generation against my velocity. Now it's just perfect. Have you been also working on the noise? it looks better, or maybe it's just the effect of faster generation. Also, the initial loading is faster by about 10-20%

I haven't noticed any difference after changing TexGen, and I was using RangeFactor 2.1, 3 and 4. 3 is good, but it's hard to say really.

I had to reduce PolyCount to 7500 to get 30 FPS, otherwise I got 25 on my Athlon 2500, Geforce 6200 256 MB video RAM, and 1 GB RAM.

Also, good work on reducing the total RAM usage. It was about 250 MB.

 

[n72.75]

could we see some pics

 

[Artlav]

I will be on holiday (if I can say...) in ten days.

Just this for today: the behavior in hover with Vcollision.dll (computerex), on the basis of the ground of Mars, is "brutal". I know that all this is not completed and that you also seek to develop a system managing the collisions. But for the hour, and whoever would be interested to know, Orulex and Vcollision does not do good housework, it seems.

That is quite expected - neither was made for the other, and if there is also meshland on, the result could be some cheap launches.

And good holidays to you!

I haven't noticed any difference after changing TexGen, and I was using RangeFactor 2.1, 3 and 4. 3 is good, but it's hard to say really.

I had to reduce PolyCount to 7500 to get 30 FPS, otherwise I got 25 on my Athlon 2500, Geforce 6200 256 MB video RAM, and 1 GB RAM.

Are you sure you wrote 7500?
Because i don't get noticeable difference with Orulex on and off when polycount is like that. Have you per chance changed some other settings?

Also, good work on reducing the total RAM usage. It was about 250 MB.

Was that on the Moon with Lv10 textures, or some other set?

could we see some pics

Hm. "Some"?

Here is Orulex-Ringworld, for example:

The current version looks pretty much the same as 1.0 in pictures, the difference is only visible in motion.

 

[Enjo]

Errata:
That was 75000 vs 95000, and it really made a 5 FPS difference. No other changes, just checked it.

The ram usage is below 150 MB, not 250, and I'm using Moon lvl8

 

[Enjo]

Atrlav, do you know that Orulex still CTDs when a ship leaves Earth? It occurs to me with lvl8 heightmap and at altitude of 12.85 Mm.

 

[Artlav]

Atrlav, do you know that Orulex still CTDs when a ship leaves Earth? It occurs to me with lvl8 heightmap and at altitude of 12.85 Mm.

Sounds bad.
Is there any certain way to reproduce it?
I tried to do a candle launch in "Cape Canaveral", and ejection burn in "DG in orbit" scenarios, no CTD in both cases all the way to half an AU.

 

[Enjo]

http://www.elwico.pl/~ender-sz/orbiter-pdf/pliki/orulex-ctd.zip

Here's a playback that stops immediately before the CTD. Keep "Play at recording speed". I got the same effects with "DG in orbit" scenario., so it looks like the playback won't help you at all.

here's a last log entry:

http://www.elwico.pl/~ender-sz/orbiter-pdf/pliki/orulex-last-log.zip

 

[Enjo]

Sorry, Artlav, but the CTDs were caused by ... my own addon (fortunately unreleased yet). I thought I was testing Orulex on a clean install, but it was not so clean.

Now I can perform a long awaited Solar Trip

 

[fort]

And good holidays to you!

What a holidays !

A "crash" of the slave HD and a long long time to try to recover the datas ( MBR destroyed or something like that, boot sector... partition table in a bad state and so on...).

I've found Disktest and PTedit...UltimateCD...

So, back to Orulex but the holidays...

Thank you Artlav.

fort

 

[Buzz Lovell]

Hi,

Since I moved to 1.00 (now 1.01) I am getting a lot of transparent terrain (see below). Any ideas?

Thanks!

http://www.photobox.ie/album/album_fullsize.html?c_photo=1708552107

 

[Buzz Lovell]

Hi

I have been doing some more work on this - it seems the surface of the moon is actually below the landscape that Orulex draws - when I fly upwards for a few kms I pass through the Orulex "surface" and everything looks great. Is there a problem with the heightmap maybe?

Thanks!

 

[jedidia]

No, I think everything's just fine. The surface has to be below the Orulex Landscape, otherwise you couldn't fly into valeys. You can also activate Meshland, which will give collision detection for the Orulex surface, but it's yet too unstable to land on it. So, for now, the only place you can land are bases.

 

[Artlav]

Since I moved to 1.00 (now 1.01) I am getting a lot of transparent terrain (see below). Any ideas?
http://www.photobox.ie/album/album_fullsize.html?c_photo=1708552107

I can only advice to try 1.1 posted above, since the link you provided requires login/password, and the description of the problem sounds unfamiliar.

I have been doing some more work on this - it seems the surface of the moon is actually below the landscape that Orulex draws - when I fly upwards for a few kms I pass through the Orulex "surface" and everything looks great. Is there a problem with the heightmap maybe?

No problems with heightmaps, just lack of collision detection and landscape awareness in Orbiter physics core. It can be helped a bit by Meshland extension, that can be found among the links above, but no proper solution yet.

 

[Buzz Lovell]

Thanks folks. Meshland is actally working great for me on the moon, can land on the Sea of Tranquility without any problem. Does Orulex take into account landing on flat areas of the moon?

Great add on! Thanks so much for all the long hours of hard work that must have gone into it!

 

[Artlav]

Thanks folks. Meshland is actally working great for me on the moon, can land on the Sea of Tranquility without any problem. Does Orulex take into account landing on flat areas of the moon?

With default settings about 40% of the Moon is flat (on the default sphere level), and it takes care of making the surface bases flat. Then, the Orbiter takes care of keeping the vessels landed on those flat spaces.

 

[jedidia]

can land on the Sea of Tranquility without any problem.

oh... strange. I allways ended up upside down. then again, I was too caught up lately to test the 1.1 Version.

 

[computerex]

Hello Artlav. Here are my system specs:
AMD Sempron 3100+ 512 MB RAM nVidia GF 6100 1.8 GHz

On a clean install of Orbiter, without the orulex/meshland plug-in, I get around 37+ FPS running on resolution 1280 x 1024. This is while looking at land, when looking at the sky I get 89+ FPS. While running FPS, I get 25 or lower FPS while looking at the sky, and around 18 20 FPS when looking at land. The attached picture is for your reference. If you are wondering why I am such a fireball, meshland+low FPS aren't a good combination. :rofl:
Everything is on default settings, in a "plug and play configuration".

Another problem, this happens with or without the alpha version, since this thread is more recent, I'll report it here. World Studio does not start for me, it crashes with a white screen. A standard, "white screen crash" (picture attached).

Pretty cool add-on Artlav, you have certainly outdone yourself.

 

[Artlav]

The most interesting thing is that this is an old thread. And i only just realized that...

Moderators, please lock this one to avoid further confusion.

The current (most recent) testing version is here:
http://orbiter-forum.com/showthread.php?p=20374

Another problem, this happens with or without the alpha version, since this thread is more recent, I'll report it here. World Studio does not start for me, it crashes with a white screen. A standard, "white screen crash" (picture attached).

Could you try the newer version on the link above, and/or also the most recent WS: http://orbides.1gb.ru/orbf/ws-080705.zip ?
Just to make sure the discussion and bugs are up-to-date.

Same to Buzz Lovell and everyone else who wanted to test Orulex since July bump of this thread.