Project Orbiter Galaxy

[ghostrunner01]

So does the texture generator just not like Nvidia? Is there any way to fix this?
I think its really strange how gas and ice planets work but not rock planets. Somethings is wrong with the texture generator.

In my opinion fixing this should be up on the priority list. If you release Orbiter Galaxy with this bug then you will get lots of angry people complaining about it not working.:facts:
Don't get me wrong though, this is an Orbiter-changing addon in this stage right now. I can't wait for the colony generator and all the other additions to this. I just think we should get this sorted out before getting to far into the development and running into more problems.

 

[jedidia]

First, I can't fix anything in the texture generator. Way above my head. All tinkering with it is done by Artlav, who is also the original author. And it's not the only project he's working on, and certainly not the one with the highest priority.

The texture generator doesn't like NVidia because Artlav has an ATI card. I.e. he can't test on NVidia cards. Still, the whole thing should work without much trouble IF the card supports all neccessary features. From what Artlav told me so far, the problem in your case and the case of Donatello is the error reporting, which is handled differently on NVidia and ATI cards.

In other words, your cards encounter an error due to a missing feature, and should in this case revert to CPU generation. It seems like it is this fallback on the CPU that doesn't work on your cards, due to unproper error handling. I.E. your cards aren't up to the task one way or the other. The only thing that can be fixed is that the card realises it and directs the task back to the CPU. That means it'll take the same time to generate the textures even if this is fixed (even a bit longer, because if you run it directly with CPU the GPU doesn't have to initialise the process and then pass it back). So there's really not much good a fix will do for you at this point. It'll be CPU generation any way you turn it, the cards just don't support the neccessary features (this thing is written for current hi-end cards, it's simply too demanding a task for anything else).

Gas giants and ice planets work because they don't need any higher shader functions whatsoever. They're the most simple of the bunch.

 

[ghostrunner01]

First, I can't fix anything in the texture generator. Way above my head. All tinkering with it is done by Artlav, who is also the original author. And it's not the only project he's working on, and certainly not the one with the highest priority.

The texture generator doesn't like NVidia because Artlav has an ATI card. I.e. he can't test on NVidia cards. Still, the whole thing should work without much trouble IF the card supports all neccessary features. From what Artlav told me so far, the problem in your case and the case of Donatello is the error reporting, which is handled differently on NVidia and ATI cards.

In other words, your cards encounter an error due to a missing feature, and should in this case revert to CPU generation. It seems like it is this fallback on the CPU that doesn't work on your cards, due to unproper error handling. I.E. your cards aren't up to the task one way or the other. The only thing that can be fixed is that the card realises it and directs the task back to the CPU. That means it'll take the same time to generate the textures even if this is fixed (even a bit longer, because if you run it directly with CPU the GPU doesn't have to initialise the process and then pass it back). So there's really not much good a fix will do for you at this point. It'll be CPU generation any way you turn it, the cards just don't support the neccessary features (this thing is written for current hi-end cards, it's simply too demanding a task for anything else).

Gas giants and ice planets work because they don't need any higher shader functions whatsoever. They're the most simple of the bunch.

Oh. :facepalm:

 

[jedidia]

Phewww, Periodic table of the elements is finished. Includes Lanthanoids and Actinoids, at least those who have a half-life of more than a few thousand years. I guess I won't need the short-lived ones, considering what I need it for... on the other hand, I'll probably need the half-life for the longer -lived ones, considering there may easily be planets too old to hold any of them. We'll see.

What the table is not currently including is abundance, but I've already found a few sources with data about that, although they're all far from complete. In the worst case, I'll just have to make a few uneducated guesses. Abundance will be variable anyways, depending on the population of a star, but an overall abundance in the galaxy would be a nice start. But I might just make a different table containing only the abundance for certain populations.

Anyways, The really tough part now will be to get a small list of the most typically occuring compounds together to form the new lithosphere/hydrosphere/atmosphere-model. Once I got a sufficiently complete list of compounds together, everything should work pretty much by itself: Create a random composition of the proto-planet based on the abundance tables, fuse the elements to compounds, check wheather they're in gaseous, liquid or solid state, and that's that. But making that compound-table WILL be a pain...

 

[Wishbone]

[ame="http://en.wikipedia.org/wiki/Abundance_of_the_chemical_elements"]Abundance of the chemical elements - Wikipedia, the free encyclopedia[/ame]

 

[jedidia]

Thanks, I knew that one, unfortunately they don't have a complete table (they have one for the solar system, though). But the more I think about it the more see that a different table with abundances for different populations would indeed be more suitable than putting the abundance directly into the periodic table...

 

[Wishbone]

I am thinking of approximating the abundance table with a quick-n-dirty statistical distribution (log-normal, or whatever fits, even Zipf). This way instead of punching in dubious numbers hundreds of times you'd enter only the parameters of the distribution. The distribs themselves are available through AlgLib.

 

[jedidia]

I am thinking of approximating the abundance table with a quick-n-dirty statistical distribution (log-normal, or whatever fits, even Zipf). This way instead of punching in dubious numbers hundreds of times you'd enter only the parameters of the distribution. The distribs themselves are available through AlgLib.

An interesting idea... I'll take a look at it.

 

[Wishbone]

There is definitely a high-frequency rollercoaster pattern in the table (linked to what isotope is stable and what isn't, I believe), but one can abstract from that for the moment. Just bear in mind that the ordinate is log-transformed itself, so it would make an entirely different picture in the absolute abundances (hugely biased towards hydrogen). If fitting and then re-normalizing (for the ratios to sum to 1) a polynomial, log abundances would be fine for non-linear least squares estimation, while absolute abundances will not.

 

[jedidia]

To tell the truth, I'm not sure I understand half of what you're saying... :lol:

I don't have any Isotopes in the table, that would be taking it a bit too far I think.

 

[donatelo200]

I just figured out why the venusian planets look crazy. It's not the pressure or where the atmosphere ends. It's the atmospheric haze! The haze on venusian planets is out of wack compared to the atmosphere as the haze for certant ones will be over 2,000km above where the atmosphere actually ends. I'd just suggest you use a similar function for the haze as you do for the altitude limit for atmospheres.

 

[Columbia42]

I recently took a trip to the Fomalhaut star system and the venusian planet there has such an incredibly oversized amount of atmospheric haze that by the time you zoom in enough for orbiter to render the atmosphere, the haze fills the entire screen.

 

[Izack]

Hmm, it seems Orbiter Galaxy doesn't generate textures for me. When the window is supposed to minimise and generate textures, it does nothing (even after half an hour of idling) and Orbiter crashes on arrival at the new system, with reports of missing planet textures.
It also creates a scenario called !OGSSwitch with the description "if you see this file, something went terribly wrong. sorry for the inconvieniance, you may delete it!" so I'm guessing something went kangaroo on me...

 

[jedidia]

I'd just suggest you use a similar function for the haze as you do for the altitude limit for atmospheres.

I thought that I do just that...? mybe a slight glitch in there somewhere, I'll have another look at it. The haze should have a direct relation to the atmosphere limit...

...and Orbiter crashes on arrival at the new system, with reports of missing planet textures.
It also creates a scenario called !OGSSwitch with the description "if you see this file, something went terribly wrong. sorry for the inconvieniance, you may delete it!" so I'm guessing something went kangaroo on me... 12-20-10 11:46 PM

Something is majorly cangoroo (what kind of expression is that? :lol... Orbiter does NOT report missing planet textures, the planet simply show up without them. Could you post the contents of your Orbiter Log file, please?

that The !OGSwitch scenario is still there only means that orbiter crashed during loading, so Orbiter Galaxy never got the chance to delete it again.

Hmm, it seems Orbiter Galaxy doesn't generate textures for me. When the window is supposed to minimise and generate textures, it does nothing (even after half an hour of idling)

Please be more specific: does the window minimize and then nothing happens, or doesn't it even minimize? If the second is the case, you're most probably trying to export a system you already exported (or tried to export). That system will be noted in the cache and not get exported again until it gets thrown out (i.e. removed from the hard drive).

To manually remove a system from cache that wasn't exported succesfully, open cache.txt in the orbitergalaxy folder and remove the corresponding system. Make sure the number in the top-most line shows the correct number of systems in the cache again after your finished.

---------- Post added at 04:59 PM ---------- Previous post was at 08:27 AM ----------

I just got around to look at the code, everything seems to be as I thought it should be. Atmospheric haze extent should always be 80% of atmospheric altitude limit...? Donatello, are you sure that the haze goes above the altitude limit (as in "confirmed by looking at the planets .cfg")? If yes, I'm completely baffeled currently.

---------- Post added at 05:24 PM ---------- Previous post was at 04:59 PM ----------

:facepalm:

Ok... I got the bugger. I mistook AtmHazeExtent for AtmHorizonAlt... the later is having absolutely ridiculous values, and it's that that's responsible for the overblown atmospheres. Will be fixed soon, but first I need to figure out what exactly AtmHazeExtent is supposed to be... "Width parameter for extent of horizon haze rendering" somehow isn't ringing a bell at all...

---------- Post added at 05:43 PM ---------- Previous post was at 05:24 PM ----------

Done. Patch is up! thanks for the awareness, Donatello!

 

[donatelo200]

No problem and yes i saw it in the config after deciding to get close to a venusian planet and launching a probe to it's surface.

 

[Wishbone]

A reminder, if you don't mind: please update the date on the add-ons you upload, otherwise it's hard to see what has been changed.

 

[jedidia]

uups, I never even noticed that I have to change the date...

I'm always renaming to the new version number and have the most actual fix at the top, but I'll see to the date in the future.

 

[jedidia]

This post goes only to Apollo13 or anyone else that might have an NVidia G 80 card or better and having the black texture bug when using GPU generation!

In short, Artlav was surprised to find that the texture generator doesn't work on the GT220, he said they should be capable of running it (indeed, anything from G 80 upwards should), suggesting that there is something wrong besides the error handling.

He put together an **EXPERIMENTAL** fix that might solve the problem, might not solve the problem or might not work at all, especially on other graphics card (if you don't run an Nvidia card and don't have problems, don't even bother!).

I'd like to ask you to download the attached zip and put the swtexgen.dll into your orbiter root directory. It's a good idea to back up the old one, so you don't have to re-download the whole orbiter galaxy in case it doesn't work at all. Please tell me if this fix changes anything, for better or worse.

 

[Linguofreak]

I'm always renaming to the new version number and have the most actual fix at the top, but I'll see to the date in the future.

Just to reduce confusion: "Actual" and "Aktuell" are false cognates. I think you mean "current" or "recent". The meaning of "actual" is closer to "real" or "true".

 

[jedidia]

Ah yes... happens to me every once in a while.