Sign in to follow this  
Guest tunger

Use of LOD values greater than 10 in FS2004

Recommended Posts

The relative merits of LOD10 vs LOD12 mesh in FS2004, when constructed from 10m source data.(I created a new thread because this topic is of some importance and not really related to the topic where this discussion began.)Link to Justin's comment re LOD]I've done some more testing. (Maybe I need an acronym for that phrase, I use it often enough :)) I compared the default mesh with 10mLOD10 and 10mLOD12 mesh for the same area of the Grand Canyon illustrated in the link in Justin's post. My screenshots can be downloaded at:www.fs-traveler.com/cgi-bin/fs9-gc-lod-test.zipJustin,1) I don't know what version of FS you are using, but my off-the-shelf version of FS9 includes LOD9 mesh for the region of the Grand Canyon illustrated in your screenshots. The filename, demgrc19[/b.dem, indicates this LOD. So my default mesh is far worse than the 10m10 or 10m12 mesh. You should verify this and perhaps update the slideshow in your link as this is very misleading. (And perhaps check some of your other slideshows as well?)2) A quick check of my LOD10 and LOD12 images suggests that the only difference is in the arrangement of the textures, but careful examination reveals a bit more.* If you examine the area where the textures are different, there are a very few locations where you can see very small changes in elevation. But these account for a very small fraction of 1% of the total area, and will never be seen while flying, however low or slowly.* There is a slight difference along the river edges, but this is due to the transition between mesh and the fixed elevation FS scenery. LOD, not mesh detail, always produces these effects.* The LOD10 mesh shows added detail all the way to the end of the canyon. When you view the LOD12 mesh, the detail in the distance reverts to the default mesh.Conclusion:The Grand Canyon is an extreme example of rugged terrian, yet even here LOD12 adds so little detail it will never be noticed while flying. And this mesh is only rendered for a radius of about 6-7 nm, or an area of about 150 square miles, and requires 22MB of data to do so.LOD10 mesh will appear identical is use, covers a radius of 30-40nm, or an area of about 2,800 square miles, and requires only 2.4MB of data.So I still don't believe there is any reasonable basis for using an LOD greater than 10 in FS2004.Stevewww.fs-traveler.com

Share this post


Link to post
Share on other sites
Help AVSIM continue to serve you!
Please donate today!

HI Steve,>>I've done some more testing. (Maybe I need an acronym for that>phrase, I use it often enough :)) How about IDSMT? :-)>I compared the default mesh>with 10mLOD10 and 10mLOD12 mesh for the same area of the Grand>Canyon illustrated in the link in Justin's post. My>screenshots can be downloaded at:>>www.fs-traveler.com/cgi-bin/fs9-gc-lod-test.zip>>Justin,>>1) I don't know what version of FS you are using, but my>off-the-shelf version of FS9 includes LOD9 mesh for the region>of the Grand Canyon illustrated in your screenshots. The>filename, demgrc19.dem, indicates this LOD. So my>default mesh is far worse than the 10m10 or 10m12 mesh. >OK, my fault there. I found an old test file in there from my Rockies terrain. I removed it and the default >is< considerably coarser.>You should verify this and perhaps update the slideshow in>your link as this is very misleading. (And perhaps check some>of your other slideshows as well?)>Yes, done so. And added the 30m source LOD10 image. So there are now four images in the sequence.http://portal.fsgenesis.net/modules.php?se...e=slideshow.php>Conclusion:>>The Grand Canyon is an extreme example of rugged terrian, yet>even here LOD12 adds so little detail it will never be noticed>while flying. And this mesh is only rendered for a radius of>about 6-7 nm, or an area of about 150 square miles, and>requires 22MB of data to do so.>Well, guess I'm going to have to disagree with you, Steve, respectfully. I can see what I would call a significant difference in the near distance, far enough to notice it. If you look at the difference between the second/third images and the fourth image, I see quite a difference on the nearby canyon walls across the way.>LOD10 mesh will appear identical is use, covers a radius of>30-40nm, or an area of about 2,800 square miles, and requires>only 2.4MB of data.What appears identical to me is the two LOD10 images whether compiled from 10m source or 30m source. The only way to tell the two images have changed in the slideshow is by looking at the captions. LOD10 compiled from 30m source looks identical to the LOD10 compiled from the 10m source. I guess this begs the question, why even use 10m source to compile LOD10m mesh if the result is identical?>>So I still don't believe there is any reasonable basis for>using an LOD greater than 10 in FS2004.>I think LOD12 mesh, compiled from 10m source, has a place, if only to provide highly-detailed terrain when on the ground or low and slow, even if its effective range is more limited than LOD10. I agree that it doesn't serve much of a purpose for the high-flying-heavy-iron drivers, but those who like to tool around in a Cub, helo, or whatnot, LOD12 is great.I just see no advantage to separating LOD10/30m source from LOD10/10m source, since they both appear to provide identical results.So I guess we can agree to disagree here.:-) -------Justinhttp://www.fsgenesis.netHigh Quality Scenery for FS200x

Share this post


Link to post
Share on other sites

Hi Justin,"OK, my fault there. I found an old test file in there from my Rockies terrain. I removed it and the default >is< considerably coarser."Good. I was concerned that Beta testers might have received an unofficial "Pro" version!I agree with your interpretation of your images, mine just don't look like that (I've also added a 30m LOD10 image). Maybe you need to go back again and make sure there are no other "old test file"(s) in your series. :)"What appears identical to me is the two LOD10 images whether compiled from 10m source or 30m source. ... I guess this begs the question, why even use 10m source to compile LOD10m mesh if the result is identical?"And can we look forward to yet another image demonstrating that 30m LOD10 mesh is no better than 90m LOD10 mesh?"So I guess we can agree to disagree here."Well, the operative word here is "we". Remember, my claim that there is no difference between 10m LOD10 and LOD12 mesh is made in support of the original post announcing this discovery. Original PostSo the preponderence of evidence favors using an LOD of 10. Your's is the only dissenting opinion, supported by version two (so far) of a set of now suspect screenshots.:)Steve

Share this post


Link to post
Share on other sites

HI Steve,>I agree with your interpretation of your images, mine just>don't look like that (I've also added a 30m LOD10 image).>Maybe you need to go back again and make sure there are no>other "old test file"(s) in your series. :)>No "old test files" left.>And can we look forward to yet another image demonstrating>that 30m LOD10 mesh is no better than 90m LOD10 mesh?>Of course not. We both know 30m LOD10 mesh is much better than 90m LOD10 mesh. And we both know why.>>So the preponderence of evidence favors using an LOD of 10.>Your's is the only dissenting opinion, supported by version>two (so far) of a set of now suspect screenshots.>So I then stand alone in my opinion that LOD12 has a place. Just a slight philosophical difference here, or a slight fs9.cfg settings difference, I don't know.But if you feel my screenshots are "suspect," so be it. As Dave Mason once said:"There ain't no good guy, there ain't no bad guy,There's only you and me and we just disagree."-------Justinhttp://www.fsgenesis.netHigh Quality Scenery for FS200x

Share this post


Link to post
Share on other sites

We have got a very detail terrain data ( 10m grid, 5m hight resolution ) for a city of Prague.When compiled in LOD10, few very thin valleys dissapeard.On the other way when compilig in LOD11 those valleys appeared, looked great.When compilingh in LOD12, the level of detail didn't grow so much, but the visible distance of detail landscape did shrink a lot.So we decided to use LOD11 as a fine balance between accurancy and possibility to see the scenery when flying.

Share this post


Link to post
Share on other sites

Curious results. I assume you are using FS2004. (Your results are typical for FS2002.)As Justin correctly noted, other cfg settings may account for some of the small differences we are seeing. "We have got a very detail terrain data ( 10m grid, 5m hight resolution ) for a city of Prague."I wonder if this data may play a role in determining what you are seeing. You must have your TMVL setting at 20 or better to see the mesh at all. But these higher settings reveal additional detail, including flaws in the data. While 10m resolution is very high, the 5m height is very poor. (Most USGS data is 1 meter, some of it is 0.1 meter.) Such low resolution elevation data usually produces mesh too "blocky" to be useful with TMVL settings above the default. So I wonder if we aren't comparing apples and oranges here.More careful testing suggests that LOD 12 mesh has a radius of only about 6nm, while LOD11 increases the radius to about 10nm. You may want to create another set of mesh at LOD9, which has a radius of about 35nm. By using both together, you should see a significant improvement in detail over a far greater area.Regards,Steve

Share this post


Link to post
Share on other sites

You are right that I have been developing it at FS2002. I changed my TMVL to 21 because of experiments with LOD11 and 12.I didn't take closer look at FS2004 ( I have enough work with AFCAD2 betatesting and VPT scenery ), but I will as soon as possible.I know that 5m resolution is poor but we use 2 ways linear interpolation that gets us rid of those so called stairs.Unfortunately there is no way for us to get better vertical resolution and stay freeware, because local geographical data are wery expensive at Czech republic.Thank you for your suggestion with LOD9 area, it is a great idea and I'll try it.Marky Parky

Share this post


Link to post
Share on other sites

Because I'm interested in viewing some regions in the highest level of detail I've been following this discussion with interest. I reciently did a comparison of my own: FSGensis LOD 10 rocky mountain region data vs my own data at LOD 12. You can see the comparison at: http://www.tumtum.com/flightsim/LodCompare.htmlMy conclusion is that I can see minor differences but not much greater detail in the LOD 12 data.

Share this post


Link to post
Share on other sites

Hi Tom,"The LOD 10 data is FSGensis's rocky mountain set. I believe that this was compiled from 10 meter resolution data, under sampled to 38.8 meter using LOD 10."I may be wrong, but I believe that mesh was constructed from 30 meter data. If so, we would expect some difference. Steve

Share this post


Link to post
Share on other sites

Hi Tom:thanks for sharing the comparison images. I doubt that anyone doing a blind test would be able to distinguish between the two LOD *in flight* - since that's what this simulator is supposed to be for ;-)Now if we could just get someone to replace those darned salt-and-pepper textures for the desert Southwest. They look great from above 10,000 feet but near ground level.... It might be interesting to try orthophoto overlays for some of those areas but I suspect that photoreal textures would share a similar fate, that is getting, 'smeared' in those vertical areas, just where we need the horizontal striations of the sandstone. Coincidentally, this is the same problem (high-contrast but 'spotty' ground textures) that, in my opinion, cancels any gain from data compiled at more than LOD10. The Southwest and similar desert areas might be bare enough to experiment with higher resolutions but if you have fairly dense autogen on top of the default textures I doubt you'd be able to actually recognize the small improvement (One other group of areas that might gain from a higher LOD might be those that have very little relief but small incised or raised features of interest).I've done trials for small watersheds in B.C. for which I have DEM data at resolutions from 10m to 250m. Obviously, I couldn't do a true blind test by myself but I couldn't even distinguish between LOD9 and LOD10 in flight, only in screenshots, and I am intimately familiar with those areas. Cheers, Holger

Share this post


Link to post
Share on other sites

> I may be wrong, but I believe that mesh was > constructed from 30 meter data. If so, we > would expect some difference. Your right. I confirmed on the web. I've updated my web page.Now I'm really, really surprised by the similarity of the terrain.Tom

Share this post


Link to post
Share on other sites

"Have you adjusted the TERRAIN_MAX_VERTEX=19 default setting in fs9"Yes. I set it to 21. I'm told higher setting shave no effect, but I may test that myself.Tom

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this