It is currently Mon, 16-09-19, 0:32 GMT

All times are UTC




Post new topic Reply to topic  [ 196 posts ]  Go to page Previous  1, 2, 3, 4, 5, 6, 7, 8 ... 14  Next
Author Message
 Post subject:
PostPosted: Mon, 24-09-07, 20:04 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4642
Location: Hamburg, Germany
ElChristou wrote:
Andrea wrote:
2- I see that as most of textures, the new one is missing the Northern ice covering...


Off topic again (sorry :oops:) but indeed this is a bit annoying... :?


It's what BMNG has published. Everyone who wants is welcome to add a layer with ice ;-)

Bye Fridger

_________________
Image


Top
 Profile  
 
 Post subject:
PostPosted: Mon, 24-09-07, 21:08 GMT 
Offline
Site Admin
User avatar

Joined: Thu, 30-08-07, 22:52 GMT
Posts: 2727
Location: France, South, not far from Montpellier
t00fri wrote:
ElChristou wrote:
Andrea wrote:
2- I see that as most of textures, the new one is missing the Northern ice covering...


Off topic again (sorry :oops:) but indeed this is a bit annoying... :?


It's what BMNG has published. Everyone who wants is welcome to add a layer with ice ;-)

Bye Fridger


Easy to say! One would have to edit the 85k raw first to be able to pass the map through your tools! I think with my 256Mo of ram my PS will have some trouble... :oops:


Top
 Profile  
 
 Post subject:
PostPosted: Mon, 24-09-07, 21:23 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4642
Location: Hamburg, Germany
ElChristou wrote:
t00fri wrote:
ElChristou wrote:
Andrea wrote:
2- I see that as most of textures, the new one is missing the Northern ice covering...


Off topic again (sorry :oops:) but indeed this is a bit annoying... :?


It's what BMNG has published. Everyone who wants is welcome to add a layer with ice ;-)

Bye Fridger


Easy to say! One would have to edit the 85k raw first to be able to pass the map through your tools! I think with my 256Mo of ram my PS will have some trouble... :oops:


Well you have not even tested the 'dwarf' version without ice, which I specially made for French people in Paraquay with only 256 Mo of RAM ;-) . So you barely have a vote about the ice. ...

I am using SCIENTIFIC data and they have been published WITHOUT the ice. So the lacking ice is NOT really MY problem... I would know a number of ways, how to implement the ice.

But this would

-- not use any authorized data about the ice layer (which is bad)
-- would be too contrived to explain to non-expert users. Clearly mounting a layer of ice by some custom tools is no big issue. I may write such a tool at some very much later stage of development...

---------------------------------------------------------
But please find me the 85k/64k layer of polar ice FIRST!! ...before complaining that it is lacking.
---------------------------------------------------------

Actually, I seem to remember that some version on Snowy does have the ice layer. But it has bathymetry as well. I don't want that either...

Bye Fridger

_________________
Image


Top
 Profile  
 
 Post subject:
PostPosted: Mon, 24-09-07, 21:40 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4642
Location: Hamburg, Germany
DW,

I remotely installed everything on my office machine, which runs a 2.8 GHz P4 and Suse 9.1. This has gcc 3.3.3. Note that the outfile1k-pre2.png has been diffed with the one from my core2Duo which gave error=0.

Applying nvimgdiff with the outfile1k-pre3.png generated on my office machine gave EXACTLY the same non-vanishing error output than my local 3.2.GHz Intel machine. So I think things are definitely not understood, yet.

Cheers,
Fridger

_________________
Image


Top
 Profile  
 
 Post subject:
PostPosted: Mon, 24-09-07, 22:04 GMT 
Offline
Site Admin
User avatar

Joined: Thu, 30-08-07, 22:52 GMT
Posts: 2727
Location: France, South, not far from Montpellier
t00fri wrote:
ElChristou wrote:
t00fri wrote:
ElChristou wrote:
Andrea wrote:
2- I see that as most of textures, the new one is missing the Northern ice covering...


Off topic again (sorry :oops:) but indeed this is a bit annoying... :?


It's what BMNG has published. Everyone who wants is welcome to add a layer with ice ;-)

Bye Fridger


Easy to say! One would have to edit the 85k raw first to be able to pass the map through your tools! I think with my 256Mo of ram my PS will have some trouble... :oops:


Well you have not even tested the 'dwarf' version without ice, which I specially made for French people in Paraquay with only 256 Mo of RAM ;-) . So you barely have a vote about the ice. ...


:shock: I beg you pardon... but:

http://www.forum.celestialmatters.org/viewtopic.php?t=45&postdays=0&postorder=asc&start=44

(it's not so old... :?)


t00fri wrote:
I am using SCIENTIFIC data and they have been published WITHOUT the ice. So the lacking ice is NOT really MY problem... I would know a number of ways, how to implement the ice.

But this would

-- not use any authorized data about the ice layer (which is bad)
-- would be too contrived to explain to non-expert users. Clearly mounting a layer of ice by some custom tools is no big issue. I may write such a tool at some very much later stage of development...

---------------------------------------------------------
But please find me the 85k/64k layer of polar ice FIRST!! ...before complaining that it is lacking.
---------------------------------------------------------


Again, I beg you pardon but perhaps you take this a bit too personal here:

- I never said it was YOUR problem,
- I wasn't complaining but saying it was annoying (I suppose we still need realistic view of earth from space, no? I'm sorry but I find annoying and not realistic a Earth without ice coverage...)

Now, I confirm it's not your fault if BMNG don't show the ice coverage. Is that fine?


Top
 Profile  
 
 Post subject:
PostPosted: Mon, 24-09-07, 22:20 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4642
Location: Hamburg, Germany
ElChristou wrote:
Fridger, I run your 3 examples with the dwarf test files without problems, but now I need to understand the logic behind those commands (remember I'm not used to commands lines), so just a couple of questions (correct me if I'm wrong). First in fact there is 2 examples here, 2) and 3) are used to obtain the VT set, no? ( 1)-> a simple png, 2) and 3)-> a set of tiles)

So what I wonder is what happen to the srtm_ramp2.world.5400x2700.bin.gz in step 2? (it is just stored in the output waiting for step 3?

Now in step 3:

Code:
3) gzip -dc < world.200406.3x5400x2700.bin.gz | tx2rgba 5400 specmap.bin | tx2pow2 4 5400 | txtiles 4 4096 1


So in a few words, take world.200406.3x5400x2700.bin.gz + the result of tx2rgba over (srtm_ramp2.world.5400x2700.bin.gz + (specmap over world.watermask.5400x2700.bin)), mix them throw a tx2pow2 then a txtiles to obtain the vt tiles...

J'ai bien compris? :oops:


Yes, essentially.

Example 2) does the following: It uses the srtm_ramp2 elevation map and finds MUCH improved ocean boundaries from the equation 'elevation=0 (<=> ocean)', as compared to the official watermask. Moreover, a little amount of rest-light (12%) is injected into the land areas, in order to achieve finer detail with the normalmap and still avoid a "plastic" look. Then the new ocean boundaries are LOGICALLY OR'ed with the existing watermask, meaning effectively that one retains the best of the two maps ;-)

The result is printed as specbin.bin and used as specmap in example3. The rest you got correctly.

Cheers,
F.

_________________
Image


Top
 Profile  
 
 Post subject:
PostPosted: Mon, 24-09-07, 23:12 GMT 
Offline
User avatar

Joined: Tue, 04-09-07, 2:32 GMT
Posts: 433
Location: South Korea
t00fri wrote:
I am without words, after spending much time to locate and trace the problem with fastfloor() and fastnint() under Linux. But, I am still not convinced that your result is what would be seen on a native 32bit Linux installation! You do not know how the magic numbers get affected with 64bit. In any case the int specs are different now from what I use. Next you installed all this on a VM...


Theoretically I don't really expect 64-bit to make a difference - after all the bit layouts of ints and doubles are the same on 32-bit and 64-bit - although the ((int*)&) had me worried so it was good to check whether it might cause problems. Anyway looking at your P4 results it's really weird indeed. I'll try and look for reasons why Linux + Pentium is not giving the results we're expecting.


Top
 Profile  
 
 Post subject:
PostPosted: Mon, 24-09-07, 23:24 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4642
Location: Hamburg, Germany
dirkpitt wrote:
t00fri wrote:
I am without words, after spending much time to locate and trace the problem with fastfloor() and fastnint() under Linux. But, I am still not convinced that your result is what would be seen on a native 32bit Linux installation! You do not know how the magic numbers get affected with 64bit. In any case the int specs are different now from what I use. Next you installed all this on a VM...


Theoretically I don't really expect 64-bit to make a difference - after all the bit layouts of ints and doubles are the same on 32-bit and 64-bit - although the ((int*)&) had me worried so it was good to check whether it might cause problems. Anyway looking at your P4 results it's really weird indeed. I'll try and look for reasons why Linux + Pentium is not giving the results we're expecting.


DW,

of course I have no clue right now. But after carefully studying that paper over the weekend,
// http://www.stereopsis.com/sree/fpu2006.html

it seemed to me that the method was a VERY clever HACK, the machine independence of which was not manifest, really. For instance, did you get a warning about

x2pow2.cpp:84: warning: dereferencing type-punned pointer will break strict-aliasing rules
tx2half.cpp:41: warning: dereferencing type-punned pointer will break strict-aliasing rules

in your new Linux installation, without applying the gcc option '-fno-strict-aliasing'? This warning is exactly due to this ((int*)&x) construct, of course. if I would replace it by ((unsigned char*)&x) for example, the warning goes away. Ignoring that warning (i.e. without adding -fno-strict-aliasing) just gives a uniformly black texture, for example.

Cheers,
Fridger

_________________
Image


Top
 Profile  
 
 Post subject:
PostPosted: Tue, 25-09-07, 0:49 GMT 
Offline
User avatar

Joined: Tue, 04-09-07, 2:32 GMT
Posts: 433
Location: South Korea
t00fri wrote:
it seemed to me that the method was a VERY clever HACK, the machine independence of which was not manifest, really.


Sree Kotay did claim that the hack is not only supposed to work on x86, but is "just as valuable on Sparc powered boxes, 68k processors, etc", i.e., presumably on any cpu with IEEE compliant fpu. Theoretically it seemed solid..

Quote:
For instance, did you get a warning about

x2pow2.cpp:84: warning: dereferencing type-punned pointer will break strict-aliasing rules
tx2half.cpp:41: warning: dereferencing type-punned pointer will break strict-aliasing rules


Yes, I indeed get these warnings. Since it's come down to this, it might be worth trying to fix this in code (as opposed to just using -fno-strict-aliasing) - and according to the following article it might be possible: "Understanding Strict Aliasing". I'll take a look at it today. I'm also trying to install OpenSUSE natively on a PIII - will get back to you with results.


Top
 Profile  
 
 Post subject:
PostPosted: Tue, 25-09-07, 16:11 GMT 
Offline
User avatar

Joined: Sun, 02-09-07, 16:06 GMT
Posts: 346
Location: Rome, Italy
t00fri wrote:
t00fri wrote:
ElChristou wrote:
Andrea wrote:
2- I see that as most of textures, the new one is missing the Northern ice overing...

Off topic again (sorry :oops:) but indeed this is a bit annoying... :?

It's what BMNG has published. Everyone who wants is welcome to add a layer with ice ;-)
Bye Fridger

I am using SCIENTIFIC data and they have been published WITHOUT the ice. So the lacking ice is NOT really MY problem... I would know a number of ways, how to implement the ice.
But this would
-- not use any authorized data about the ice layer (which is bad)
-- would be too contrived to explain to non-expert users. Clearly mounting a layer of ice by some custom tools is no big issue. I may write such a tool at some very much later stage of development...
---------------------------------------------------------
But please find me the 85k/64k layer of polar ice FIRST!! ...before complaining that it is lacking.
---------------------------------------------------------
Bye Fridger

Fridger, just like ElChristou, I know very well that the images we use to apply your nmtools and F-Textools are not yours, so obviously our complaint on missing ice was not directed to you, but to the producers of those images, so this sentence
---------------------------------------------------------
But please find me the 85k/64k layer of polar ice FIRST!! ...before complaining that it is lacking.
---------------------------------------------------------

is not correct, no one is complaining with you! :wink:
And, just to explain a bit more my thought, I don't understand the reason why they produced 12 monthly maps, where ALL the other seasonal variations are well shown, but all of them without North Pole ice.
This looks a bit strange. :shock:
The only thing I could suppose was the lack of polar satellites capable to image poles at such a resolution, but this is not the case, because the ice shelves in Antarctica on those 12 monthly maps change extension month after month, so surely both poles have been imaged correctly. :cry:
I would like to understand the reason of the ice missing, just for curiosity.
You say, correctly, "I am using SCIENTIFIC data and they have been published WITHOUT the ice", but I think that this doesn't mean that such missing data cannot be added, using scientific data from other fonts.
I mean, if there is an existing set of the Arctic polar cap, at 8k, 16k or 32k resolution, taken on a given and known date, why not add this polar cap to the the (same month, obviously!) 86400x43200 map? 8)
In this way the only difference should be the cap resolution, but the overall result would be the real one, i.e. what astronauts on board of ISS are looking just now.
What do you think to be more “correctâ€ÂÂ

_________________
Andrea
WorldPoss Sn Team Member
Core 2 Quad Q6600 G0 3.6 GHz- 4 GB DDR2- DELL 2709W 1920x1200-WinXP 32 SP3- VISTA 64- ASUS P5K-E- 8800 GTX 768MB- SATA II 320-300 GB-IDE 1 TB- 169.21- Celestia1.4.1_patch3- 1.5.1-LUA Edu Tools


Top
 Profile  
 
 Post subject:
PostPosted: Tue, 25-09-07, 16:55 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4642
Location: Hamburg, Germany
Andrea,

apparently BOTH Christophe and you have misinterpreted my above line:

Quote:
---------------------------------------------------------
But please find me the 85k/64k layer of polar ice FIRST!! ...before complaining that it is lacking.
---------------------------------------------------------


What is behind is the following.

In order not to clash with the strict Celestia policy of documentation of sources, we need a clean scientific source for a hires polar ice texture. Only then am I ready to think about a simple (automatic) process to implement such a layer. I know there is a bad low-resolution one in some tiff version of BlueMarble, for example. But I will NOT support such bad graphics to ruin the nice textures generated with the F-TexTools.

All these considerations were behind my above request. I also did not feel attacked personally by Christophe. Vraiment, chèr ami ;-) Neither was my above response meant to read that Chris was complaining at my address! It was beyond doubt, however, that Christophe complained (in general terms ) about the unsatisfactory situation concerning the polar ice layer.

So really my above line was not meant to be agressive in any way.

Bye Fridger

PS: Andrea, I think it would be very good for the future readability of this thread, if you could somewhat reduce the width of your two Earth textures!? Many thanks.

_________________
Image


Last edited by t00fri on Tue, 25-09-07, 17:40 GMT, edited 2 times in total.

Top
 Profile  
 
 Post subject:
PostPosted: Tue, 25-09-07, 16:59 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4642
Location: Hamburg, Germany
Another most plausible reason for the missing polar ice in the BMNG texture is of course this one ;-)

BMNG = Blue Marble Next Generation

So this just means they display the world as it will look to the Next Generation, when apparently the polar ice will have melted away ;-) . I am sure you are all aware about the fierce discussions about that issue worldwide...

Cheers,
Fridger

_________________
Image


Top
 Profile  
 
 Post subject:
PostPosted: Tue, 25-09-07, 17:25 GMT 
Offline
User avatar

Joined: Sun, 02-09-07, 16:06 GMT
Posts: 346
Location: Rome, Italy
t00fri wrote:
Andrea,apparently BOTH Christophe and you have misinterpreted my above line:
Quote:
---------------------------------------------------------
But please find me the 85k/64k layer of polar ice FIRST!! ...before complaining that it is lacking.
---------------------------------------------------------


What is behind is the following.
Without clashing with the strict Celestia policy of documentation of sources, we need a clean scientific source for a hires polar ice texture. Only then am I ready to think about a simple (automatic) process to implent such a layer. I know there is a bad low-resolution one in some tiff version of BlueMarble, for example. But I will NOT support such bad graphics to ruin the nice textures generated with the F-TexTools.
All these considerations were behind my above request. I also did not feel attacked personally by Christophe. Vraiment, chèr ami ;-) Neither was my above response meant to read that Chris was complaining at my address! It was beyond doubt, however, that Christophe complained (in general terms ) about the unsatisfactory situation concerning the polar ice layer.
So really my above line was not meant to be agressive in any way.
Bye Fridger
PS: Andrea, I think it would be very good for the future readability of this thread, if you could somewhat reduce the width of your two Earth textures!? Many thanks.

OK Fridger, all clear now, we misunderstood each other. :wink:
But, are we sure that there is nothing "acceptable" for the Northen Polar cap?
I think that for "normal" (non-scientific) use no one goes to search for that iceberg ot the other one :lol: , so perhaps a lower resolution map like Blue Marble could be sufficiently satisfactory, or not?
Believe me, when I show the Earth to my students, I find embarassing to explain them that there is no northern ice cap, due to "scientific reasons" so, to avoid this, I'm compelled to show the Jmii's map North Polar cap, with all the pinches there!
Anyhow, we can live even without it, obviously. :wink:
Regarding the size of the double Earth image, just reduced.
Bye

Andrea :D

_________________
Andrea
WorldPoss Sn Team Member
Core 2 Quad Q6600 G0 3.6 GHz- 4 GB DDR2- DELL 2709W 1920x1200-WinXP 32 SP3- VISTA 64- ASUS P5K-E- 8800 GTX 768MB- SATA II 320-300 GB-IDE 1 TB- 169.21- Celestia1.4.1_patch3- 1.5.1-LUA Edu Tools


Top
 Profile  
 
 Post subject:
PostPosted: Tue, 25-09-07, 18:05 GMT 
Offline
User avatar

Joined: Tue, 04-09-07, 2:32 GMT
Posts: 433
Location: South Korea
Finally, with some more googling and the help of a 32-bit OpenSUSE installation, I was able to track down the bug!

It's indeed a Linux/glibc x86-specific issue, and is basically due to Linux setting an unorthodox FPU precision mode.
Even better, there is documented code that appears to fix the issue. Look here: Resolution of Differences Between X86 Linux/glibc Floating-Point to Integer Conversion Behavior Relative to Other Platforms

Long story cut short: It's an x86(32-bit) glibc "feature".
That's correct, it's a feature, not a bug according to GNU. Like, whatever. Rounding 1.500 to 1, and 1.499 to 1.5 is just wrong. :roll:

The optimized tx2pow2 and tx2half are confirmed to work properly on 32-bit Linux after applying this simple fix.
Fridger: I've sent you a mail with the modifications I made.


Top
 Profile  
 
 Post subject:
PostPosted: Tue, 25-09-07, 18:28 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4642
Location: Hamburg, Germany
dirkpitt wrote:
Finally, with some more googling and the help of a 32-bit OpenSUSE installation, I was able to track down the bug!

It's indeed a Linux/glibc x86-specific issue, and is basically due to Linux setting an unorthodox FPU precision mode.
Even better, there is documented code that appears to fix the issue. Look here: Resolution of Differences Between X86 Linux/glibc Floating-Point to Integer Conversion Behavior Relative to Other Platforms

Long story cut short: it's not a cpu bug, it's an x86(32-bit) glibc "feature".
That's correct, it's a feature, not a bug according to GNU. Like, whatever. :roll:

The optimized tx2pow2 and tx2half are confirmed to work properly on 32-bit Linux after applying this simple fix.
Fridger: I've sent you a mail with the modifications I made.


DW,

just got home from work... What a joy to read your mail and the resolution of our rather persistent Linux problem.

So, the original fastfloor() and fastnint() were indeed a very clever HACK, in the sense that Linux on a x86 (32bit) machine stole the show ;-) . Glibc unfortunately has quite a number of special "glitches" ...

I'll look into your mods right away. Your email has also arrived already. Many thanks.

Now what do you think: I would be inclined to post a .pre3 version ASAP. I think it would make the code much more transparent if we commit ourselves to only the fastfloor() and fastnint() solutions that are now known to work on all three OS. Also I think we can substitute your two-pass version in tx2pow2 for good. I checked it in various environments with nvimgdiff and it always agreed with the one-pass. What do you think??

Eventually, I also would like to substitute it in the nmtools, including fastfloor and fastnint!
Cheers,
Fridger

_________________
Image


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 196 posts ]  Go to page Previous  1, 2, 3, 4, 5, 6, 7, 8 ... 14  Next

All times are UTC


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group